JP5463494B2 - Technology trend information generator - Google Patents

Description

  The present invention is, for example, a related term acquisition device that acquires a related term related to a certain term, or a technical trend information that creates technical trend information that is information that associates a technical term with a period related to a technical document including the technical term. It relates to a creation device.

  A conventional information processing apparatus searches a Web page on the Web and automatically collects related terms (see, for example, Non-Patent Document 1, Non-Patent Document 2, Non-Patent Document 3, and Non-Patent Document 4). In general, in order to collect a term related to a certain technical term t from the Web, a procedure of first collecting descriptions related to the term t and extracting a term related to t therefrom is required. Here, one of the points is a method of collecting an appropriate description regarding the term t. For example, in Non-Patent Document 1, descriptions relating to the term t are collected by the method described below. For a certain term t, first, four types of queries “t”, “t”, “t” and “t” are input to the search engine, and the top 100 pages of the obtained URL are obtained. To do. Next, the obtained page is shaped and divided into sentences, only sentences including the term t are extracted, and related terms are collected therefrom.

In the prior art, there is a survey paper creation support system that takes into account reference information between papers (Non-Patent Document 5).
In recent years, research on automatically collecting terms related to a certain term from a text set has been actively performed (see Non-Patent Documents 1, 3, 4, and 6).

Risa Sato, 1 other, "Automatic collection of related terms using the web", IPSJ Research Report, Natural Language Processing, NL-153, p. 57-64, 2003 Yasuhiro Sasaki, two others, “Proposal of an index for measuring the degree of association between terms”, 10th Annual Conference of the Association for Natural Language Processing, p. 25-28, 2004 Kiyoaki Shirai and three others, “Attempts to Create Portal Sites Automatically”, 10th Annual Conference of the Association for Natural Language Processing, p. 624-627, 2004 Yusuke Ohara and three others, “Gathering Related Terms Using the Web”, 3rd Information Science and Technology Forum (FIT 2004), 2004 Eizo Namba and 1 other, “Development of survey paper creation support system considering reference information between papers”, Natural Language Processing, Vol. 6, no. 5, p. 43-62, 1999 Eiwa Namba, “Automatic Collection of Related Terms Using Citation Information Between Papers”, 11th Annual Conference of the Association for Natural Language Processing, 2005

  However, the conventional information processing apparatus does not focus on the citation relationship between documents, and collects related terms for each term by collecting related pages using a Web search engine and extracting related terms therefrom. There was a problem that it took time. Further, the conventional information processing apparatus has a problem that the accuracy of collecting related terms is low because it does not pay attention to the citation relationship of documents.

  In addition, it is indispensable to collect and organize information such as “what kind of technology has been used since” in a research field in order to overview research trends in that field. However, there has been a problem that collecting such information requires a great deal of labor and time for a person familiar with the research field.

  Further, in the above-mentioned non-patent documents and the like, as described above, terms related to a certain term can be automatically collected. However, in the above non-patent documents and the like, it has not been possible to create information indicating when technical terms have been used.

  The present invention has been made to solve at least one of the above problems, and provides a related term acquisition device and the like that can acquire appropriate related terms by using a citation relationship between documents. With the goal.

  In addition, the present invention has been made to solve at least one of the above-described problems, and it is possible to create technical trend information that is information associating a technical term with a period related to a technical document including the technical term. An object of the present invention is to provide a technical trend information creation device and the like.

  In order to achieve at least one of the above objects, a related term acquisition apparatus according to the present invention receives a reference information storage unit that stores two or more reference information that is information indicating a reference, and term information that is information indicating a term. Search the document information having the term indicated by the term information received by the term information receiving unit and the term information received by the term information receiving unit from all or a part of the document information from the two or more document information stored in the document information storage unit. , Having at least cited document identification information that is information for identifying another document cited in the retrieved document information, which is the entire retrieved document information or a part of the retrieved document information The document information acquisition unit that acquires the part from the document information storage unit and the cited document identification information included in all or part of the document information acquired by the document information acquisition unit Cited reference information, which is information indicating the contribution, is searched from two or more reference information stored in the reference information storage unit, and corresponds to a predetermined item that is a part of the searched reference information. Related to the term indicated by the term information received by the term information receiving unit from a part of the cited document information acquired by the cited document information acquiring unit and the cited document information acquiring unit that acquires the portion to be acquired from the document information storage unit The related term information acquisition part which acquires the related term information which is the information which shows related term, and the related term information output part which outputs the related term information which the said related term information acquisition part acquired.

  With such a configuration, related term information related to term information can be acquired. In particular, when the cited document information acquisition unit acquires a part of the cited document information, more appropriate related term information can be acquired.

  In the related term acquisition device according to the present invention, the document information includes structure description information that is information describing a structure of the document information, and the cited document information acquisition unit is acquired by the document information acquisition unit. A part of cited document information indicating a cited document identified by the cited document identification information included in all or a part of the document information, and is associated with a predetermined item by the structure description information. Part may be acquired.

  With such a configuration, since the document information includes the structure description information, the cited document information acquisition unit can easily acquire a portion corresponding to a predetermined item.

  The related term acquisition device according to the present invention further includes a cited document structure information acquisition unit that acquires cited document structure information that is information indicating correspondence between one or more parts of the cited document information and items of the part. The cited document information acquisition unit acquires a part of the cited document information corresponding to a predetermined item using the cited document structure information acquired by the cited document structure information acquisition unit. May be.

  With such a configuration, the cited document information acquisition unit can easily acquire a portion corresponding to a predetermined item using the cited document structure information acquired by the cited document structure information acquisition unit.

  In the related vocabulary acquisition apparatus according to the present invention, the document information acquisition unit, when acquiring a part of the document information, is a part having at least the cited document identification information, and is identified by the cited document identification information. The cited document identification information included in all or part of the document information acquired by the document information acquisition unit is acquired. It is a part of cited document information that is information indicating the document identified by, and is determined in advance according to the citation format in the citation information included in all or part of the document information acquired by the document information acquisition unit You may acquire the part corresponding to the item.

  With such a configuration, the cited document information acquisition unit can acquire the part corresponding to the item according to the citation format, and acquires a part of the cited document information corresponding to which item according to the citation format. By appropriately setting whether or not, it is possible to obtain more appropriate related term information.

In the related term acquisition device according to the present invention, the related term information acquisition unit acquires related term candidate information that is information that is a candidate for a related term in a part of the cited reference information acquired by the cited reference information acquisition unit. Related term candidate information acquiring means, a related degree calculating means for calculating the degree of association of terms indicated by the related term candidate information acquired by the related term candidate information acquiring means for each related term candidate information, and the related term candidate information Related term information determining means for determining related term candidate information having a higher corresponding degree of related term information as related term information among the related term candidate information acquired by the acquiring unit may be provided.
With such a configuration, related term information can be appropriately acquired from a part of the cited document information acquired by the cited document information acquisition unit.

  Moreover, in the related vocabulary acquisition apparatus by this invention, in the said literature information storage part, 1st type literature information and 2nd type literature information are memorize | stored at least, The said literature information acquisition part is said literature information. All or part of the first type of document information is acquired from the first type of document information stored in the storage unit, and the cited document information acquisition unit is stored in the document information storage unit. A part of the cited document information may be acquired from the second type document information.

  With such a configuration, it is possible to obtain related terms related to a certain term and related terms in a field different from the term. For example, patent terms related to terms used in academic literature can be obtained.

  In order to achieve at least one of the above objects, the technical trend information creation device according to the present invention includes a plurality of document information that is information indicating at least a part of a technical document, and a period related to the technical document indicated by the document information. A document information storage unit that stores at least time information that is information to be displayed, and a document that acquires all or a part of at least some document information among a plurality of document information stored in the document information storage unit Corresponding to an information acquisition unit, a technical term extraction unit that extracts technical terms from all or part of the reference information acquired by the reference information acquisition unit, and reference information including the technical terms extracted by the technical term extraction unit Using a time information acquisition unit that acquires time information from the document information storage unit, a technical term extracted by the technical term extraction unit, and a time information acquired by the time information acquisition unit A technical trend information creating unit that creates technical trend information, which is information that associates technical terms with a period related to technical literature including the technical terms, and an output unit that outputs the technical trend information created by the technical trend information creating unit; , With.

  With such a configuration, it is possible to create technical trend information that is information for associating a technical term with a period related to a technical document including the technical term. As a result, by using the created technology trend information, it is possible to easily know the transition of technology used in a certain research field. In addition, there is an advantage that it is not necessary to spend a great deal of labor and time for those who are familiar with the research field in order to create the technology trend information.

The technical trend information creation device according to the present invention further includes a search term receiving unit that receives a search term, and a search unit that searches the document information storage unit using the search term received by the search term receiving unit. The document information acquisition unit may acquire all or part of the retrieved document information.
With such a configuration, it is possible to specify a research field in which technology trend information is created by a search term.

  In the technical trend information creation device according to the present invention, a search term storage unit storing two or more search terms, and any one of the two or more search terms stored in the search term storage unit A search term selection accepting unit that accepts the selection, and the search term accepting unit may accept the search term indicated by the selection accepted by the search term selection accepting unit.

  With such a configuration, it is possible to avoid a situation in which an inappropriate search term is input by storing an appropriate term (for example, a term indicating a research field) as a search term in the search term storage unit. It can be done. As a result, an appropriate search can be performed and appropriate technical trend information can be created.

Further, in the technical trend information creation device according to the present invention, a technical term extraction unit that extracts technical terms from the literature information stored in the literature information storage unit, a suffix storage unit that stores a predetermined suffix, When a term obtained by adding a suffix stored in the suffix storage unit to the technical term extracted by the technical term extraction unit is included in the document information stored in the document information storage unit, the technical term And a search term storage unit that stores the search term as a search term in the search term storage unit.
With such a configuration, the search terms can be automatically stored in the search term storage unit without the user performing processing for storing the search terms in the search term storage unit.

In the technical trend information creation device according to the present invention, the search unit may search for document information corresponding to a technical document whose search term received by the search term receiving unit is a main topic.
With such a configuration, it is possible to create technical trend information in a field whose search term is a main topic.

  The technical trend information creation device according to the present invention further includes a specifying unit that specifies document information having the same field as the document information searched by the search unit, and the document information acquiring unit is specified by the specifying unit. All or part of the document information may be acquired from the document information storage unit.

  With such a configuration, although it is document information that cannot be acquired by a search using a search term, other document information having the same field as the document information acquired by the search can be acquired. As a result, it is considered that more literature information can be collected and more appropriate technical trend information can be created.

Further, in the technical trend information creation device according to the present invention, the specifying unit is cited document identification information which is information for identifying another technical document cited in the document information from the document information searched by the search unit. And using the cited document identification information, document information having the same field as the document information searched by the search unit may be specified.
With such a configuration, it is possible to specify document information using citation relationships.

  Further, in the technical trend information creation device according to the present invention, the specifying unit is citation information that is information indicating a description of another technical document cited in the document information from the document information searched by the search unit. If the citation format indicated by the citation information is a problem-pointing citation format that points out the problem of the cited technical document, the technical document described in the citation information is identified. The cited document identification information may be used to identify document information having the same field as the document information searched by the search unit.

  With such a configuration, it is possible to specify document information using citation information. In particular, when the citation format is a problem-pointing citation format that points out the problem of the cited technical document, the fields of the cited document information and the cited document information are similar. Is known by the inventors' research, and appropriate literature information can be specified.

In the technical trend information creation device according to the present invention, the technical term extraction unit includes a clue word storage unit that stores one or more clue words, and all or a part of the document information acquired by the document information acquisition unit. A clue phrase searching unit for searching for the clue phrase and a technical term extracting unit for extracting a term having a predetermined relationship with the clue phrase searched by the clue phrase searching unit as a technical term.
With such a configuration, it is possible to extract technical terms from all or part of the acquired document information using the clue words / phrases.

In the technical trend information creation device according to the present invention, the technical term extraction unit includes a clue word storage unit that stores one or more clue words, and all or a part of the document information acquired by the document information acquisition unit. , A technical term for extracting as a technical term a clue phrase search means for searching for the clue phrase, a search term received by the search term acceptance unit, and a term associated with the clue phrase searched by the clue phrase search means Extraction means.
With such a configuration, technical terms can be extracted from all or part of the acquired document information using the clue phrases and the search terms.

In the technical trend information creation device according to the present invention, the technical term extraction unit includes a clue word storage unit that stores one or more clue words, and all or a part of the document information acquired by the document information acquisition unit. A clue word / phrase search means for searching for the clue word / phrase, a technical term storage means for storing two or more technical terms, and a term having a predetermined relationship with the clue word / phrase searched by the clue word / phrase search means. When the technical term memorize | stored in the memory | storage means is contained, you may provide the technical term extraction means which extracts the said technical term as a technical term.
With such a configuration, it can be avoided that the technical term extraction means extracts general nouns, non-technical terms and the like as technical terms.

In the technical trend information creation device according to the present invention, the technical term extraction unit includes a clue word storage unit that stores one or more clue words, and all or a part of the document information acquired by the document information acquisition unit. A clue phrase searching means for searching for the clue phrase, a technical term candidate extracting means for extracting a term having a predetermined relationship with the clue phrase searched by the clue phrase search means, and the technical term candidate extraction Technical term selection means for selecting all or part of technical term candidates extracted by the means as technical terms.
With such a configuration, it is possible to first extract technical term candidates using the clue words and then select an appropriate technical term from the technical term candidates.

  In the technical trend information creation device according to the present invention, the technical term extraction unit further includes technical term suffix storage means for storing one or more suffixes that can be taken by the technical term, and the technical term selection means includes: A technical term candidate having a suffix stored in the technical term suffix storage means may be selected as a technical term.

  With such a configuration, a term having a predetermined suffix is considered to be appropriate as a technical term. By selecting a technical term using the suffix, an appropriate technical term can be selected. It can be done.

  In the technical trend information creation device according to the present invention, the technical term selection means selects a technical term candidate used as a main topic in any document information stored in the document information storage unit as a technical term. May be.

  With such a configuration, a term that is not used as a main topic in any document information is considered to be unsuitable as a technical term, but a technical term candidate used as a main topic in any document information is technically used. By selecting as a term, an appropriate technical term can be selected.

In the technical trend information creation device according to the present invention, the technical term extraction unit includes literature information from which the technical term candidates are extracted, literature information identification means for identifying literature information in the same field, and the technical terms. A technical term suffix storage unit that stores one or more possible suffixes, and the document information acquisition unit also acquires all or part of the document information specified by the document information specifying unit, The technical vocabulary selection means includes all or a part of one document information acquired by the document information acquisition unit and the document information specified by the document information specifying means that the field is the same as the one document information. When at least one of the technical term candidates extracted from all or part of the technical term candidates has the suffix stored in the suffix storage means, it is extracted from all or part of the one document information. It may select technical term candidates as technical terms.
With such a configuration, it is possible to select an appropriate technical term.

Further, in the technical trend information creation device according to the present invention, the technical trend information is an image indicating a timing indicated by timing information corresponding to document information including the technical term for each technical term, and the output unit includes the output Technology trend information may be displayed and output.
With such a configuration, the technology trend can be easily grasped by the user viewing the displayed image of the technology trend information.

  In the technical trend information creation device according to the present invention, the technical trend information indicates, for each technical term, a reference graphic corresponding to the reference information including the technical term is indicated by time information corresponding to the reference information. It is an image that is displayed according to time, a graphic selection receiving unit that accepts selection of a literature graphic in the technical trend information, and information related to literature information corresponding to the literature graphic that the graphic selection receiving unit has accepted selection, A display information acquisition unit that acquires display information that is information to be displayed on the image of the technical trend information from the literature information storage unit, and the technical trend information creation unit displays the display acquired by the display information acquisition unit You may create the technical trend information which displays information.

  With such a configuration, for example, the user viewing the technical trend information image can easily know the predetermined information about the document information including the technical term displayed in the technical trend information image. As a result, the convenience of the user is improved.

  According to the related term acquisition device or the like according to the present invention, related term information related to term information can be acquired. In addition, for example, it is possible to acquire more appropriate related term information by acquiring the portion corresponding to the item corresponding to the citation format and acquiring the related term information.

  According to the technical trend information creation device and the like according to the present invention, it is possible to create technical trend information that is information that associates technical terms with the time related to technical literature including the technical terms. Therefore, by using the technology trend information, it is possible to easily know the transition of the technology used in a certain research field.

The block diagram which shows the structure of the related vocabulary acquisition apparatus by Embodiment 1 of this invention. The block diagram which shows the structure of the related vocabulary information acquisition part by the embodiment The flowchart which shows operation | movement of the related vocabulary acquisition apparatus by the embodiment The flowchart which shows operation | movement of the related vocabulary acquisition apparatus by the embodiment The flowchart which shows operation | movement of the related vocabulary acquisition apparatus by the embodiment The flowchart which shows operation | movement of the related vocabulary acquisition apparatus by the embodiment The figure for demonstrating the example of an experiment in the embodiment The figure for demonstrating the example of an experiment in the embodiment The block diagram which shows the structure of the related vocabulary acquisition apparatus by Embodiment 2 of this invention. The block diagram which shows the structure of the related vocabulary information acquisition part by the embodiment The block diagram which shows the structure of the related vocabulary information acquisition part by the embodiment The flowchart which shows operation | movement of the related vocabulary information acquisition part by Embodiment 3 of this invention. The figure which shows an example of the clue phrase dictionary for type C in the embodiment The figure which shows an example of the clue phrase dictionary for type B in the embodiment The figure which shows an example of the bibliographic information of the literature information in the embodiment The block diagram which shows the structure of the technical trend information preparation apparatus by Embodiment 4 of this invention. The flowchart which shows operation | movement of the technical trend information preparation apparatus by the embodiment The flowchart which shows operation | movement of the technical trend information preparation apparatus by the embodiment The flowchart which shows operation | movement of the technical trend information preparation apparatus by the embodiment The flowchart which shows operation | movement of the technical trend information preparation apparatus by the embodiment The flowchart which shows operation | movement of the technical trend information preparation apparatus by the embodiment The block diagram which shows the structure of the technical vocabulary extraction part in the embodiment The flowchart which shows operation | movement of the technical vocabulary extraction part in the embodiment The block diagram which shows the structure of the technical vocabulary extraction part in the embodiment The flowchart which shows operation | movement of the technical vocabulary extraction part in the embodiment The block diagram which shows the structure of the technical vocabulary extraction part in the embodiment The flowchart which shows operation | movement of the technical vocabulary extraction part in the embodiment The block diagram which shows the structure of the technical vocabulary extraction part in the embodiment The flowchart which shows operation | movement of the technical vocabulary extraction part in the embodiment The block diagram which shows the structure of the technical vocabulary extraction part in the embodiment The flowchart which shows operation | movement of the technical vocabulary extraction part in the embodiment The flowchart which shows operation | movement of the technical vocabulary extraction part in the embodiment The figure which shows an example of a response | compatibility with the structure tag and clue word phrase in the embodiment The figure which shows an example of the information memorize | stored in the literature information storage part in the embodiment The figure which shows an example of a response | compatibility with the position of the clue phrase of type C and the clue phrase in the embodiment The figure which shows an example of the literature information which the literature information acquisition part in the embodiment acquired The figure which shows an example of the clue word phrase memorize | stored in the clue word phrase memory | storage means in the embodiment The figure which shows an example of the technical vocabulary extracted in the embodiment The figure which shows an example of the acquired time information in the embodiment The figure which shows an example of the technical trend information output in the same embodiment The figure which shows an example of the technical trend information in the embodiment The figure which shows an example of the technical term memorize | stored in the technical term memory | storage means in the embodiment The block diagram which shows the structure of the technical trend information creation apparatus by Embodiment 5 of this invention. The flowchart which shows operation | movement of the technical trend information preparation apparatus by the embodiment The flowchart which shows operation | movement of the technical trend information preparation apparatus by the embodiment The block diagram which shows the structure of the technical trend information creation apparatus by Embodiment 6 of this invention. The flowchart which shows operation | movement of the technical trend information preparation apparatus by the embodiment The figure which shows an example of the technical trend information output in the same embodiment Figure showing experimental results Figure showing experimental results Figure showing experimental results The block diagram which shows the structure of the technical trend information preparation apparatus by another form. The figure which shows an example of the classification result information in another form The figure which shows an example of the technical trend information output in other forms Schematic diagram showing an example of the appearance of a computer system The figure which shows an example of a structure of a computer system

  Hereinafter, a related term acquisition device and a technical trend information creation device according to the present invention will be described using embodiments. In the following embodiments, components and steps denoted by the same reference numerals are the same or equivalent, and repetitive description may be omitted.

(Embodiment 1)
A related term acquisition apparatus according to Embodiment 1 of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a configuration of a related term acquisition device 1 according to this embodiment. In FIG. 1, the related term acquisition device 1 according to the present embodiment includes a document information storage unit 11, a term information reception unit 12, a document information acquisition unit 13, a cited document information acquisition unit 14, and a related term information acquisition unit. 15 and a related term information output unit 16.

  The document information storage unit 11 stores two or more document information. Here, the document information is information indicating a document. The document may be any document as long as it has a certain unit such as a paper, a patent document (for example, an application publication gazette, a patent gazette, etc.), a web page, an official gazette, and a book. One document information includes information (data) of a corresponding one document. However, one piece of document information may only include information in a range necessary for later processing. For example, in the case of literature information of an academic paper, when only the reference literature and title of the academic paper are used in the subsequent processing, the literature information stored in the literature information storage unit 11 includes only the reference literature and title. It may be information. The document information may be a text file, or may be a file of a unique format such as word processing software. When the document information is a text file, the text file may be described in a markup language that describes a document using tags, or the content of the corresponding document is simply indicated by text data. Also good. The document information has a structure in at least a part thereof. Here, the structure may be specified by a tag, a heading, or the like, or may not be specified. In the latter case, for example, by reading a paragraph in the document indicated by the document information, it is found that the summary is described there, and by reading the other paragraph, the purpose is described there It can be understood. When the structure is specified, the structure is information included in the document information, and may or may not be indicated by the structure description information that is information describing the structure of the document information. Here, the structure description information is, for example, a heading tag used in a Japanese patent specification, a tag used in an HTML file, or the like. As a case where the structure of the document information is described other than the structure description information, for example, as in a normal paper, there may be a case where the text data of a headline such as “summary” or “reference document” is included in the document information. is there. The two or more document information stored in the document information storage unit 11 includes at least document information indicating a document that cites another document. Other documents cited in certain document information are referred to as cited documents. The document information indicating a document that cites another document includes cited document identification information that is information for identifying the cited document. Here, the content of the cited document identification information is not limited as long as it is information that can identify the cited document. The cited document identification information may be, for example, an ID for identifying a cited document such as a patent number, an application publication number of a patent application, or a PMID (PubMed Identifier), and relates to an author name, title, date of writing, and medium It may be bibliographic information of a document that can identify a cited document such as information (magazine name, publisher name, URL, etc.). The document information storage unit 11 can be realized by a predetermined recording medium (for example, a semiconductor memory, a magnetic disk, an optical disk, etc.). The process in which the user attribute correspondence information is stored in the document information storage unit 11 does not matter. For example, document information may be stored in the document information storage unit 11 via a recording medium, and document information transmitted via a communication line or the like is stored in the document information storage unit 11. Alternatively, the document information input via the input device may be stored in the document information storage unit 11.

  The term information receiving unit 12 receives term information. Here, the term information is information indicating a term. This term information is generally information indicating the text of a term, but may be other than that as long as it is information indicating a term. In the related term acquisition device 1, a related term that is a term related to the term indicated by the term information received by the term information receiving unit 12 is acquired. For example, the term information reception unit 12 may receive information input from an input device (for example, a keyboard, a mouse, a touch panel, etc.), or may receive information transmitted via a wired or wireless communication line. Alternatively, information read from a predetermined recording medium (for example, an optical disk, a magnetic disk, a semiconductor memory, etc.) may be received. The term information receiving unit 12 may or may not include a device (for example, a modem or a network card) for receiving. The term information receiving unit 12 may be realized by hardware, or may be realized by software such as a driver that drives a predetermined device.

  The term information receiving unit 12 may receive information other than the term information. For example, when performing some kind of search or the like based on the related term information acquired by the related term acquisition device 1, a search key used in the search (for example, an international patent classification used in the search) may be received together with the term information. Even in such a case, processing such as acquisition of the following document information is performed using the term information received by the term information receiving unit 12.

  The document information acquisition unit 13 retrieves document information having the term indicated by the term information received by the term information reception unit 12 in all or part of the document information from two or more document information stored in the document information storage unit 11. To do. Document information having a term in all document information means that the term is included in any part of the document information. On the other hand, the document information having a term as part of the document information means that the term is included in a part of the document information, for example, a predetermined region such as a title or a claim. . When the document information acquisition unit 13 retrieves document information having the term indicated by the term information as a part of the document information, for example, information indicating an item corresponding to the part of the document information is stored in a recording medium (not illustrated). Thus, the literature information acquisition unit 13 may read the information indicating the item from the recording medium and perform the above search.

  The document information acquisition unit 13 acquires all of the retrieved document information or a part of the retrieved document information from the document information storage unit 11. When the document information acquisition unit 13 acquires a part of the retrieved document information, a part of the document information is information that identifies other documents cited in the retrieved document information. It is a part having at least document identification information. Therefore, all or part of the document information including at least the cited document identification information is acquired by the document information acquisition unit 13. When the document information acquisition unit 13 acquires a part of the document information, for example, the document information acquisition unit 13 includes a column of a reference document (Reference) that is considered to include at least cited text identification information, a patent document, You may acquire the column of nonpatent literature, or you may specify the area | region where cited reference identification information exists using techniques, such as pattern matching, and may acquire a part of literature information including the area | region. . In the former case, for example, information indicating an item corresponding to the part to be acquired is stored in a recording medium (not shown), and the literature information acquisition unit 13 reads the information indicating the item from the recording medium, and the item A part of the document information corresponding to may be acquired. In the latter case, for example, information indicating a pattern used for pattern matching or the like is stored in a recording medium (not shown), and the literature information acquisition unit 13 reads the pattern from the recording medium and performs pattern matching or the like. A part of the document information may be acquired by performing the above. When the document information acquisition unit 13 searches for document information having the term indicated by the term information as part of the document information, and when the document information acquisition unit 13 acquires a part of the searched document information, The part of the document information used in the search and the part of the document information to be acquired may be the same part of the document information, or may be a part of the document information different.

  The cited document information acquisition unit 14 searches the cited document information from two or more document information stored in the document information storage unit 11. Here, the cited document information is document information indicating a document identified by the cited document identification information included in all or part of the document information acquired by the document information acquisition unit 13. The cited document information acquisition unit 14 acquires a part of the retrieved cited document information from the document information storage unit 11. A part of the cited document information acquired by the cited document information acquiring unit 14 is a part of the cited document information corresponding to a predetermined item. A part corresponding to a predetermined item is a part described for the item. An item is an area that can be classified in a document. For example, a title, a title of the invention, an abstract (abstract), a claim, a claim, a claim, a preamble of a claim (preamble), and first described Claims (hereinafter referred to as “first claim”), claims subordinate to the first claim, the object of the invention, the effects of the invention, and the like. The predetermined item may be a single item or two or more items. The item may be manifested in the cited document information or may not be manifested. The fact that an item is actualized means that a portion corresponding to the item is clearly indicated in the cited document information. For example, the item may be manifested by displaying the item name in text, or the item may be manifested by using a tag. The fact that the item is not actualized means that the part corresponding to the item is not clearly shown in the cited document information as in the case where the item is actualized. For example, a case where a document is described without setting items is applicable. Even in such a case, the cited document information acquisition unit 14 can know what items are described in a predetermined part of the document.

  Here, the process in which the cited document information acquisition part 14 acquires the part of the cited document information corresponding to the predetermined item is demonstrated. When information indicating a predetermined item is stored in a recording medium (not shown) accessible by the cited document information acquiring unit 14, the cited document information acquiring unit 14 accesses the recording medium and selects the item. Read the indicated information. And when the item is actualized in cited reference information, the cited reference information acquisition part 14 acquires the cited reference information part corresponding to the read item. On the other hand, when the item is not manifested in the cited document information, the cited document information acquisition unit 14 specifies the cited document information part corresponding to the item, and acquires the specified cited document information part. Next, identification of the cited document information corresponding to the item will be described. For example, the cited document information acquisition unit 14 holds condition information indicating conditions corresponding to each item to be specified in a recording medium (not shown). When the cited document information acquisition unit 14 specifies a cited document information portion corresponding to a certain item, the cited document information acquisition unit 14 acquires condition information corresponding to the item, and obtains the cited document information portion corresponding to the condition information. Identify.

  When the structure information is included in the document information, the cited document information acquisition unit 14 is identified by the cited document identification information included in all or part of the document information acquired by the document information acquisition unit 13. A part of cited document information indicating a document, which is associated with a predetermined item may be acquired by structure description information.

  Moreover, if the cited reference information acquisition part 14 can acquire a part of above cited reference information as a result, the process will not be ask | required. For example, when the document information acquisition unit 13 acquires information including the title of the cited document as the cited document identification information, and the cited document information acquisition unit 14 acquires a part of the cited document information corresponding to the item “title”. The cited document information acquisition unit 14 may acquire a part of the cited document information by acquiring a part corresponding to the title of the cited document information from the cited document identification information acquired by the document information acquisition unit 13. .

  The related term information acquisition unit 15 acquires related term information from a part of the cited document information acquired by the cited document information acquisition unit 14. Here, the related term information is information indicating related terms related to the term indicated by the term information received by the term information receiving unit 12. The detailed configuration of the related term information acquisition unit 15 will be described later with reference to FIG.

  The related term information output unit 16 outputs the related term information acquired by the related term information acquisition unit 15. Here, the output may be, for example, display on a display device (for example, a CRT or a liquid crystal display), transmission via a communication line to a predetermined device, printing by a printer, or audio output by a speaker. However, it may be stored in a recording medium. The related term information output unit 16 may or may not include an output device (for example, a display device or a printer). The related term information output unit 16 may be realized by hardware, or may be realized by software such as a driver that drives these devices.

  FIG. 2 is a block diagram illustrating a configuration of the related term information acquisition unit 15. In FIG. 2, the related term information acquisition unit 15 includes a related term candidate information acquisition unit 21, a related degree calculation unit 22, and a related term information determination unit 23.

  The related term candidate information acquisition unit 21 obtains the citation obtained by the cited document information acquisition unit 14 as related term candidate information which is information as a candidate for a related term in a part of the cited document information acquired by the cited document information acquisition unit 14. Acquired from a part of literature information. The related term candidate information acquisition unit 21 may store the acquired related term candidate information in a recording medium (not shown). Here, various methods for acquiring the related term candidate information will be described here.

[Method 1 of obtaining related term candidate information]
The related term candidate information acquisition unit 21 performs morphological analysis on a part of the cited reference information acquired by the cited reference information acquisition unit 14, and acquires information indicating a single noun and / or compound noun as related term candidate information. May be. The related term candidate information acquisition unit 21 may acquire information indicating an unknown word as related term candidate information in addition to information indicating a single noun or the like. Here, the morpheme analysis is a process of dividing a natural language document into morphemes and recognizing parts of speech of the morphemes. As software for performing morphological analysis, “ChaSen” (http://chasen.naist.jp) developed at Nara Institute of Science and Technology is known in Japan. By using a tea bowl, for example, information indicating nouns can be acquired. Here, the nouns are simple nouns, unknown words, or compound nouns in which these are connected. In addition, examples of software that gives parts of speech to English words include “TnT” (http://www.coli.uni-saarland.de/˜thorsten/tnt/) and “Brill Tagger” (http: //www.cs. .Jhu.edu / ˜brill /) and the like are known. The method for acquiring information indicating a single noun or the like as related term candidate information is disclosed in Non-Patent Document 4 and the like and has been known in the art, and thus detailed description thereof is omitted.

[Method 2 of obtaining related term candidate information]
The related term candidate information acquisition unit 21 may acquire information indicating technical terms as related term candidate information from a part of the cited reference information acquired by the cited reference information acquisition unit 14. Here, a method for acquiring information indicating technical terms will be described. First, the related term candidate information acquisition unit 21 indicates a single noun or the like from a part of the cited reference information acquired by the cited reference information acquisition unit 14 in the same manner as the “related term candidate information acquisition method 1” described above. Extract information. Then, scoring indicating the possibility of being a technical term is performed for each single noun and the like, and information indicating a single noun or the like given a high score is acquired as related term candidate information. Here, a plurality of types are known as methods for assigning scores.

(Scoring based on word-making ability)
The word-making ability is a scale indicating the ease of composing a single noun compound word. A high score is given to information indicating a single noun including a noun with high word-making ability. Here, the connection frequency may be used as an index indicating the word-making ability. The concatenation frequency is the number of different words connected to a certain word and the number of types of words appearing immediately before or after a certain word. That is, the articulation frequency is a scale indicating how many words are easily included in a compound noun, in other words, how easy it is to compose a compound noun. In this way, a score can be given to each single noun or the like based on the coined word ability. In compound nouns, a geometric mean of the scores of single nouns that are constituent elements may be given as the score of the compound noun.

(Scoring based on appearance frequency)
Let t be a single noun or the like of interest. The number of constituent words of t is n. In that case, the score corresponding to t is calculated as follows.
score of t = (appearance frequency of t) / (average appearance frequency of single nouns etc. where the number of words is n)

  Here, “single noun etc.” in the above formula is a single noun etc. extracted by the related term candidate information acquisition means 21. Here, the appearance frequency of t and the average appearance frequency of a single noun having n as the number of words are calculated in a part of the cited document information acquired by the cited document information acquiring unit 14. Note that “t appearance frequency” may be simply used as the score of t.

  In addition, scoring may be performed using a method other than the above description, and information indicating technical terms may be acquired as related term candidate information using a method other than the above description. The method of acquiring information indicating technical terms as related term candidate information is disclosed in Non-Patent Document 1, Non-Patent Document 3, and the following documents, and has been conventionally known. Detailed description is omitted.

  Literature: Hiroshi Nakagawa, “Automatic Term Recognition based on Statistics of Compound Nouns”, Terminology, Vol. 6, no. 2, p. 195-210, 2000

Literature: Hirokazu Ohata, Hiroshi Nakagawa, “Extraction of technical terms by the number of words connected differently”, Information Processing Society of Japan Research Report, 2000-NL-136, p. 119-126
References: Hiroshi Nakagawa, Masanori Mori, Yasuaki Yumoto, “Extraction of technical terms based on appearance frequency and connection frequency”, natural language processing, Vol. 10 No. 1, p. 27-45, January 2003

[Method 3 of obtaining related term candidate information]
The related term candidate information acquisition unit 21 analyzes a part of the structure of the cited document information acquired by the cited document information acquisition unit 14 to identify a group of nouns, and finds a single noun included in the identified group of nouns. The information shown may be acquired as related term candidate information. Here, the group of nouns is a group including one or more single nouns. For example, it is assumed that a part of the cited document information acquired by the cited document information acquiring unit 14 is a claim of a patent application document as shown below.

  “In a working machine that drives an actuator by an operation means to perform a desired work, a load detection means for detecting a load created in the work mechanism of the work and a signal of a frequency corresponding to a detection value of the load detection means are output. A first frequency converter, a second frequency converter for outputting a pulse having a frequency corresponding to a detection value of the load detecting means, and a signal output from the first frequency converter as the second frequency converter. A virtual vibration for operating a work machine, characterized by comprising a modulation means for intermittently outputting only a pulse output period from the frequency converter and a vibration generating means for generating a vibration in accordance with the output of the modulation means Generator. ]

  The claims generally have the following structural features: The first feature is that there is a “group of nouns” at the end of the claim. In the above example, “virtual vibration generating device for operating a work machine” corresponds to the group of nouns. Secondly, when the claim is divided into the first half and the second half by a character string such as “in” or “and”, immediately before “in” or “in” Is also characterized by the existence of a “group of nouns” similar to that at the end. In the above example, “work machine” corresponds to a group of the nouns. Therefore, the related term candidate information acquisition means 21 acquires a group of nouns at the end of the claim and a group of immediately preceding nouns such as “in”, and indicates single nouns included in the group of these nouns. Information may be acquired as related term candidate information. The related term candidate information acquisition means 21 can specify the location of a group of immediately preceding nouns such as “in” and “in” as a clue phrase. Although claims of patent application documents have been described here, a group of nouns can be specified in a similar manner in documents with other structures, and related term candidate information can be acquired. .

  In addition, a method for analyzing a document structure, specifying a group of nouns, and acquiring information indicating a single noun included in the specified group of nouns is disclosed in the following documents, and has been disclosed in the past. Therefore, detailed description thereof is omitted.

References: Akihiro Niimori, Manabu Okumura, Yuzo Marukawa, Makoto Iwayama, “Structural analysis of patent claims using clue phrases”, Transactions of Information Processing Society of Japan, Vol. 45, no. 3, p. 891-905, 2004 Document: Japanese Patent Application Laid-Open No. 2004-164054.

  The related term candidate information acquisition unit 21 may exclude information indicating common nouns that frequently appear from the related term candidate information. For example, the related term candidate information acquisition unit 21 holds one or more pieces of information indicating nouns to be excluded in a recording medium (not shown), and indicates the nouns to be excluded in the information acquired as the related term candidate information. When any of the information is included, the information may be excluded from the related term candidate information. Examples of nouns to be excluded are methods, descriptions, publications, literatures, and the like.

  The degree-of-association calculating unit 22 calculates the degree of association of terms indicated by the related term candidate information acquired by the related term candidate information acquiring unit 21 for each related term candidate information. The degree-of-association calculating unit 22 may store the calculated degree of association in association with the related term candidate information acquired by the related term candidate information acquiring unit 21 in a recording medium (not shown).

  Here, a method for calculating the degree of association will be described. The degree-of-relevance calculating means 22 indicates the degree of relevance of the related term candidate information, and the term indicated by the related term candidate information is a part of the cited reference information acquired by the cited reference information acquisition unit 14 (a part of the cited reference information). If one or more is acquired, it may be the frequency of appearing in the set). In addition, the degree-of-relevance calculation unit 22 calculates, for each related term candidate information, a score indicating the possibility of being a technical term described in “Related Term Candidate Information Acquisition Method 2”, for example, a score based on word-making ability, appearance frequency Or a score based on the connection frequency may be calculated, and the score may be used as the relevance level. In this case, when the score is already calculated in the related term candidate information acquisition unit 21, the relevance degree calculation unit 22 may acquire the score from the related term candidate information acquisition unit 21, or A new score may be calculated. Accordingly, the calculation of the degree of association by the degree-of-relevance calculation unit 22 includes acquisition of the degree of association. The degree-of-association calculating unit 22 may calculate the degree of association using both the appearance frequency of the term indicated by the related term candidate information and the score. For example, the degree of association may be calculated by adding the two, or the degree of association may be calculated by multiplying the two.

  The related term information determining unit 23 determines related term candidate information having a higher related degree as related term information among the related term candidate information acquired by the related term candidate information acquiring unit 21. For example, the related term information determining unit 23 has a certain number of related terms (for example, 30) from the top of the related term candidate information acquired by the related term candidate information acquiring unit 21 or a certain ratio (for example, (30% etc.) related term candidate information may be determined as related term information, and related term candidate information whose corresponding degree of association is a predetermined threshold value or more may be determined as related term information, or other According to the method, related term candidate information having a higher corresponding degree of association may be determined as related term information. More specifically, the related term information determination unit 23 sorts the relevance level calculated by the relevance level calculation unit 22, selects related term candidate information corresponding to a higher rank of the sorted relevance level, and records it (not shown). The related term information is determined by accumulating the related term information on the medium. The related term information stored in the recording medium (not shown) is the related term information output by the related term information output unit 16. The related term information determining means 23 may store the related term information in a recording medium (not shown). In this case, the stored related term information is read and output by the related term information output unit 16.

  Next, operation | movement of the related vocabulary acquisition apparatus 1 by this Embodiment is demonstrated using a flowchart. FIG. 3 is a flowchart showing the overall operation of the related term acquisition device 1.

  (Step S101) The term information receiving unit 12 determines whether term information has been received. If the term information is accepted, the process proceeds to step S102. If not, the process of step S101 is repeated until the term information is accepted.

  (Step S <b> 102) The document information acquisition unit 13 stores two or more document information that is stored in the document information storage unit 11 with document information having the term indicated by the term information received by the term information receiving unit 12 in all or part of the document information. A cited document that is searched from the document information and is information that identifies all of the retrieved document information or a part of the retrieved document information and identifies other documents cited in the retrieved document information. A part having at least identification information is acquired from the document information storage unit 11. Details of this processing will be described later with reference to the flowchart of FIG.

  (Step S <b> 103) The cited document information obtaining unit 14 obtains cited document information that is information indicating a document identified by cited document identification information included in all or part of the document information obtained by the document information obtaining unit 13. A search is performed from two or more pieces of document information stored in the storage unit 11, and a part of the retrieved cited document information that corresponds to a predetermined item is acquired from the document information storage unit 11. Details of this processing will be described later with reference to the flowchart of FIG.

  (Step S104) The related term information acquisition unit 15 indicates information related to a term related to the term indicated by the term information received by the term information reception unit 12 from a part of the cited reference information acquired by the cited reference information acquisition unit 14. Get related term information. Details of this processing will be described later with reference to the flowchart of FIG.

(Step S105) The related term information output unit 16 outputs the related term information acquired by the related term information acquisition unit 15. Then, the process returns to step S101.
In the flowchart of FIG. 3, the process ends when the power is turned off or the process is terminated.

FIG. 4 is a flowchart showing details of the document information acquisition process in step S102 in the flowchart of FIG.
(Step S <b> 201) The document information acquisition unit 13 searches for document information including all or part of the term information received by the term information receiving unit 12 from two or more document information stored in the document information storage unit 11. .

(Step S202) The literature information acquisition unit 13 sets the counter i to 1.
(Step S203) The document information acquisition unit 13 determines whether or not the i-th document information searched in step S201 exists. If it exists, the process proceeds to step S204. If it does not exist, the series of processing ends, and the process proceeds to step S103 in the flowchart of FIG.

(Step S204) The document information acquisition unit 13 acquires all or part of the searched i-th document information. When a part of the searched i-th document information is acquired, as described above, a part including at least the cited document identification information is acquired. Note that in the searched i-th document information, when the cited document identification information is not included at all, the document information acquisition unit 13 does not need to acquire anything.
(Step S205) The literature information acquisition unit 13 increments the counter i by 1, and returns to Step S203.

FIG. 5 is a flowchart showing details of the cited document information acquisition process in step S103 in the flowchart of FIG.
(Step S301) The cited document information acquisition unit 14 sets the counter i to 1.

  (Step S302) The cited document information acquisition unit 14 determines whether or not the i-th document information acquired by the document information acquisition unit 13 exists. If it exists, the process proceeds to step S303. If it does not exist, the series of processing ends, and the process proceeds to step S104 in the flowchart of FIG.

(Step S303) The cited document information acquisition unit 14 acquires all the cited document identification information included in the i-th document information acquired by the document information acquisition unit 13.
(Step S304) The cited document information acquisition unit 14 sets the counter j to 1.

(Step S305) The cited document information acquisition unit 14 determines whether or not the j-th cited document identification information acquired in step S303 exists. If it exists, the process proceeds to step S307. If it does not exist, the process proceeds to step S306.
(Step S306) The cited document information acquisition unit 14 increments the counter i by one. Then, the process returns to step S302.

  (Step S307) The cited document information acquisition unit 14 searches the document information storage unit 11 for cited document information identified by the j-th cited document identification information acquired in step S303, and obtains the retrieved cited document information in advance. The determined part is acquired from the document information storage unit 11.

  (Step S308) The cited document information acquisition unit 14 determines whether a part of the cited document information acquired in step S307 has already been acquired. If it has already been acquired, the process proceeds to step S310. If not, the process proceeds to step S309.

(Step S309) The cited document information acquisition unit 14 temporarily stores a part of the cited document information acquired in step S307. A part of the cited document information that is additionally written in step S309 becomes a part of the cited document information finally acquired by the cited document information acquiring unit 14.
(Step S310) The cited document information acquisition unit 14 increments the counter j by 1. Then, the process returns to step S305.

FIG. 6 is a flowchart showing details of the related term information acquisition processing in step S104 in the flowchart of FIG.
(Step S401) The related term candidate information acquisition unit 21 obtains a part of the cited document information acquired by the cited document information acquisition unit 14 (if two or more of the cited document information are acquired, the cited document). Related term candidate information is acquired from a part of the cited document information temporarily stored in step S309.

(Step S402) The relevance calculation means 22 sets the counter i to 1.
(Step S403) The degree-of-association calculating unit 22 determines whether or not the i-th related term candidate information acquired from a part of the cited document information by the related term candidate information acquiring unit 21 exists. If it exists, the process proceeds to step S404. If it does not exist, the process proceeds to step S407.

  (Step S404) The degree-of-association calculating unit 22 calculates the degree of association of the i-th related term candidate information acquired from a part of the cited document information by the related term candidate information acquiring unit 21.

  (Step S405) The related term information determination unit 23 temporarily stores the degree of association calculated in step S404 in association with the corresponding related term candidate information. Here, the related term candidate information corresponding to the degree of relevance calculated in step S404 is the i-th related term candidate information acquired from a part of the cited document information by the related term candidate information acquiring unit 21. .

(Step S406) The relevance calculation means 22 increments the counter i by 1. Then, the process returns to step S403.
(Step S407) The related term information determination means 23 sorts the relevance level temporarily stored in step S405.

  (Step S408) The related term information determining means 23 determines related term candidate information corresponding to the higher related degree sorted in step S407 as related term information. Then, the series of processing ends, and the process proceeds to step S105 in the flowchart of FIG.

  Next, experimental examples showing specific effects obtained by the related term acquisition device 1 according to the present embodiment will be described. In this experimental example, it is assumed that the document information acquired by the document information acquisition unit 13 is an academic paper. On the other hand, it is assumed that the cited document information acquired by the cited document information acquisition unit 14 is document information related to a patent. Therefore, by inputting the term information indicating the academic paper term to the related term acquisition device 1, the related term information indicating the patent term related to the academic paper term is output. Here, academic paper terms are terms that are relatively frequently used in academic papers (for example, specific subordinate terms such as DRAM and word processor), and patent terms are terms that are relatively frequently used in patent documents. (For example, an abstract upper term such as a semiconductor storage device or a document editing device).

  In this experimental example, the information of the patent publication (1993 to 2002) is included in the document information stored in the document information storage unit 11. Further, the bibliographic information of about 85,000 bibliographic information of cited papers in the patent publication is also included in the bibliographic information stored in the bibliographic information storage unit 11. As a method of extracting bibliographic information of a paper cited in the patent publication from the patent publication, a method disclosed in the following document was used.

  References: Natsumi Anzen, Hideaki Namba, Teruaki Aizawa, Manabu Okumura, “Construction of a Search Environment that Integrates Patents and Article Databases”, Information Processing Society of Japan Research Report, NL-168, p. 21-26, 2005

  First, term information received by the term information receiving unit 12 will be described. Nouns are extracted from the information of the patent publications stored in the document information storage unit 11. Then, the extracted nouns are arranged in order of frequency, and the 25 paper terms are selected manually from those with the highest frequency. Specifically, “CPU”, “semiconductor laser”, “DRAM”, “memory cell”, “word processor”, “novolak resin”, “CD” and the like were selected.

  In this experimental example, the pattern of the portion of the cited document information acquired by the cited document information acquiring unit 14 is as follows.

  (Pattern 1) Information indicating the first claim of the cited document, that is, information indicating the claim described in the first of the claims is acquired as a part of the cited document information.

  (Pattern 2) All the claims of the cited document, that is, information indicating all claims described in the claims is acquired as a part of the cited document information.

  (Pattern 3) Information indicating the first claim of the cited document and information indicating the dependent claim which is a claim subordinate to the first claim are acquired as part of the cited document information. Here, claims dependent on the first claim include not only claims directly dependent on the first claim but also claims dependent on the first claim through other claims. .

  In this experimental example, the pattern in which the related term candidate information acquisition unit 21 acquires the related term candidate information is as follows.

  (Pattern A) As explained in the method 1 for acquiring related term candidate information, morphological analysis is performed on a part of the cited document information, and nouns are acquired as related term candidates.

  (Pattern B) As described in the method 3 for acquiring related term candidate information, a group of nouns is specified from a part of cited document information, and nouns included in the group of nouns are acquired as related term candidates.

  It is assumed that the related term candidate information acquisition unit 21 does not acquire a related term candidate that matches a phrase included in the unnecessary phrase list by using a predetermined unnecessary phrase list. Terms such as “method” and “description” appear in many claims regardless of the field, so that such terms are excluded from the related term candidates. In the unnecessary word / phrase list, nouns included in the patent publications for 10 years are arranged in the order of appearance frequency, and the words / phrases to be included in the unnecessary word / phrase list are selected manually from those having the highest frequency. For example, a total of 350 unnecessary words / phrases such as method, description, publication, literature, use, use, details, manufacturing, proposal, manufacturing method, data, etc. are included in the unnecessary word / phrase list.

  Further, in this experimental example, in order to verify the effect of acquiring a part of the cited document information, the relevance calculation means 22 does not calculate the relevance. The related term information determining unit 23 determines the related term candidate information acquired by the related term candidate information acquiring unit 21 as the related term information as it is.

  Under such circumstances, the term information receiving unit 12 receives each of the 25 pieces of term information described above, and a process is executed until the related term information output unit 16 outputs related term information related to the term information. The related term information output unit 16 accumulates related term information in a predetermined recording medium. After that, the related term information was used for verification.

In this verification, whether the related term information is correct or not was determined according to the following criteria.
(Criteria 1) Only the term that is conceptually closest is correct. For example, for the term information “word processor”, the related term information “document editing device” is correct, and the related term information “display device” that is a component of the word processor is not correct. The answer was correct.

(Criteria 2) Frequency of appearance in patent database When the frequency of occurrence of a certain term is extremely low in the patent database, the term is considered to be inaccurate because it is not useful for acquiring related terms of the patent.

(Criteria 3) Comparison with the one selected in Criteria 1 Even if a term does not meet the above (Criteria 2), it is judged that the term is conceptually almost equal to the one selected in (Criteria 1) above. The answer was correct even at low frequency. For example, for the term information “word processor”, the related term information “document creation device” conceptually substantially equivalent to the related term information “document editing device” is also correct. Regarding fluctuations of notation such as “laser” and “laser”, if one related term information is determined to be correct, the other is also correct.

  In this verification, the following evaluation scale ε was used. This is an extension of MRR (mean reprocratic rank) generally used in the evaluation of question answering systems, and is disclosed in the following document.

Literature: Yoji Kiyota, Ikuo Kurohashi, Toko Kido, “Resolving Dependency Displacement Using Automatically Extracted Metaphor Expressions”, Natural Language Processing, Vol. 11, no. 4, p. 127-145, 2004

  Here, n is the number of correct answers for the input. R is a set of correct answer rank numbers in the output list. ε takes a maximum value of 1 when all correct answers are ranked at the top.

  FIG. 7 is a table showing the results of this experimental example. In FIG. 7, ID is set to identify each experimental example. The experimental example of ID6 is an experimental example serving as a baseline in the other experimental examples, in which nouns that co-occur with the term information at a high frequency are extracted and the nouns are used as related terms. The experimental example of ID6 serving as the baseline is a method used in acquiring related terms from the past. As can be seen from the table in FIG. 7, when comparing the experimental example of ID1 and the experimental example of ID2, it is better not only to acquire a noun but also to acquire a noun from a group of nouns in the claims. Recognize. Moreover, when comparing the experimental example of ID3 and the experimental example of ID4, it is better not only to acquire a noun but also to acquire a noun from a group of nouns in the claims. From this, it is understood that it is effective not only to acquire a noun but also to acquire a noun from a group of nouns in the claims.

  Next, it is examined whether it is effective to consider the relationship between claims. Comparing the experimental results of IDs 2, 4, and 5, the experimental result of ID5 using the first claim and its dependent claims is the best, and the experimental result of ID2 using only the first claim is the worst result. became. The reason why the experimental result of ID2 is bad is that the restriction on the part from which a part of the cited document is acquired is too strong, not only noise is reduced, but also the number of correct answers that can be acquired is reduced. The ID3 experiment example using all the claims has a larger number of correct answers than the ID2 experiment example, but a large number of incorrect answers have been extracted, resulting in inferior results to the ID5 experiment example. It has become.

  In each method, when extracting nouns from one patent publication gazette, the number of claims used was examined, and the result shown in FIG. 8 was obtained. The ID5 experiment example has a higher accuracy than the ID4 experiment example even though the number of claims to be acquired is less than 40% compared to the ID4 experiment example. It can be seen that the correct answer is included in the term and its dependent claims with high probability.

  As described above, according to the related term acquisition device 1 according to the present embodiment, it is possible to acquire effective related term information by acquiring a part of the cited document information. Therefore, the related term acquisition device 1 that can appropriately acquire the related term information related to the accepted term information can be provided.

  The cited document information acquired by the cited document information acquiring unit 14 may be any part as long as the related term information can be appropriately acquired. For example, the cited document information is information indicating a patent document, and the portion of the cited document information corresponding to the predetermined item acquired by the cited document information acquisition unit 14 may be a part of an independent claim. . An independent claim is a claim that is not dependent on other claims. The cited document information acquisition unit 14 may determine whether a certain claim is an independent claim, for example, as follows. The cited document information acquisition unit 14 determines whether or not there is a description referring to another claim, that is, the description of “Claim N” in the claim of interest. It is determined that the current claim is not an independent claim, and if it does not exist, it is determined that the current claim is an independent claim. Here, “N” is an arbitrary integer of 1 or more. Further, for example, the cited document information is information indicating a patent document, and the cited document information part corresponding to a predetermined item acquired by the cited document information acquiring unit 14 describes the effect of the invention. It may be a part. The portion describing the effect of the invention may be, for example, a portion corresponding to the item “effect of the invention” or a portion where it is determined from the contents of the document that the effect of the invention is described. Good. In the latter case, for example, as a method for specifying the portion describing the effect of the invention, it starts with “as above” or “like this”, and “effect”, “possible”, “can” There is a method of specifying a paragraph including any of the above words as a part describing the effect of the invention. The cited document information is information indicating a patent document, and the cited document information part corresponding to the predetermined item acquired by the cited document information acquisition unit 14 is a part other than the premise part of the claims. May be. The premise part of a claim is a part called a so-called preamble of the claim, and is generally a part described before “in” or “being” in the claim. Therefore, the part other than the premise part of the claim is a part where the characteristic description of the claim is made, and is generally described after “in” or “being” in the claim. It is a part. When the preamble is not described in the claim, the entire claim is a part other than the premise part.

(Embodiment 2)
A related term acquisition apparatus according to Embodiment 2 of the present invention will be described with reference to the drawings. The related term acquisition device according to the present embodiment further includes a cited document structure information acquisition unit.

  FIG. 9 is a block diagram showing the configuration of the related term acquisition device 2 according to this embodiment. In FIG. 9, the related term acquisition device 2 according to the present embodiment includes a document information storage unit 11, a term information reception unit 12, a document information acquisition unit 13, a cited document information acquisition unit 14, and a related term information acquisition unit. 15, a related term information output unit 16, and a cited document structure information acquisition unit 31. Configurations and operations other than the cited document structure information acquisition unit 31 are the same as those in the first embodiment, and a description thereof will be omitted.

  The cited document structure information acquisition unit 31 acquires cited document structure information that is information indicating correspondence between one or more parts of the cited document information stored in the document information storage unit 11 and items of the parts. If the cited document structure information is information that can associate an item with the portion of the cited document information corresponding to the item, the structure of the information is not limited. For example, the cited document structure information may indicate an item by a tag and a portion corresponding to the item. That is, in the cited document information, a region sandwiched between <effect> and </ effect> may be a portion corresponding to the item “effect”. In this case, the cited document structure information may be stored in the document information storage unit 11. Further, for example, the cited document structure information may include information indicating the portion of the cited document information corresponding to the item “effect” in association with the item “effect”. In this case, the cited document structure information may be held in association with the cited document information in the document information storage unit 11, or a record (not shown) included in the cited document structure information acquisition unit 31 and other components. It may be held in a medium. In the latter case, it is preferable to store the cited document structure information in association with the cited document identification information. The information indicating the cited document information part may be, for example, information indicating the position of the leading end and the position of the trailing end of the cited document information part. The positions of the leading edge and the trailing edge may be indicated by the number of bytes from the leading edge of the cited document information, or may be indicated by the number of pages and the number of lines, for example.

  Here, how the cited document structure information acquisition unit 31 specifies a part corresponding to an item will be described. For example, the cited document structure information acquisition unit 31 holds condition information indicating conditions corresponding to each item to be specified in a recording medium (not shown). Then, the portion of the cited document information corresponding to the condition information is specified, and the cited document structure information is acquired by associating the item with the portion. For example, the condition information corresponding to the item “effect” starts with a paragraph starting with “as above” or “like this”, and the word “effect”, “possible”, or “can” is included in the paragraph. Suppose that it is included. Then, the cited document structure information acquisition part 31 can acquire the cited document structure information by specifying the paragraph which meets the conditions and associating the paragraph with the item “effect”. In addition, for example, the condition information corresponding to the item “characteristic description of claim” is a part included in the claim, which is a part following “is” or “in”, and a punctuation mark “. ”. Then, the cited document structure information acquisition unit 31 can acquire the cited document structure information by specifying a portion that satisfies the condition and associating the portion with the item “characteristic description of claim”. As for the method of obtaining the characteristic description of claims, the above-mentioned literature “Akihiro Niimori, Manabu Okumura, Yuzo Marukawa, Makoto Iwayama,“ Structural analysis of patent claims using clue phrases ”, Information Processing Society of Japan Journal, Vol. 45, no. 3, p. 891-905, 2004 ”.

  Next, the overall operation of the related term acquisition apparatus 2 according to the present embodiment will be described using the flowchart of FIG. The processes other than steps S501 and S502 are the same as those in the flowchart of FIG. 3 of the first embodiment, and the description thereof is omitted. However, in step S103, the cited document information acquisition unit 14 uses the cited document structure information acquired by the cited document structure information acquisition unit 31 and corresponds to a predetermined item that is part of the cited document information. The part to be acquired shall be acquired.

  (Step S501) The cited document structure information acquisition unit 31 determines whether or not the cited document structure information has already been acquired. If it has not been acquired, the process proceeds to step S502. If it has already been acquired, the process proceeds to step S102.

  (Step S502) The cited document structure information acquisition unit 31 acquires the cited document structure information for all of the cited document information stored in the document information storage unit 11. Here, among the document information stored in the document information storage unit 11, which is cited document information is unknown until the document information is acquired. It is assumed that cited document structure information is acquired for all document information stored in the storage unit 11.

  In the flowchart of FIG. 10, the case where the cited document structure information is acquired collectively has been described. However, a case where a part of the cited document information is acquired, and the cited document structure information of the cited document information is not yet obtained. If it is not acquired, only the cited document structure information of the cited document information may be acquired. In this way, unnecessary cited document structure information is not acquired, and the processing load is reduced.

  The cited reference structure information may be acquired in advance when the related term acquisition device 2 does not perform related term acquisition processing, that is, when the processing load on the related term acquisition device 2 is not large.

  Thus, according to the related vocabulary acquisition apparatus 2 according to the present embodiment, a part of the cited document information is acquired using the cited document structure information, and therefore the structure of the document is not revealed in the cited document information. Even if it exists, acquisition of the desired part of a cited reference can be easily performed by using cited reference structure information.

(Embodiment 3)
A related term acquisition apparatus according to Embodiment 3 of the present invention will be described with reference to the drawings. The related term acquisition device according to the present embodiment acquires a part of cited document information corresponding to an item corresponding to the cited format of the cited document in the document information.

  The configuration of the related term acquisition apparatus according to the present embodiment is the same as that shown in FIGS. 1 and 2 of the first embodiment, and the description thereof is omitted. In addition, when acquiring a part of literature information, the literature information acquisition part 13 is a part which has cited reference identification information at least, Comprising: The part which has at least cited information. The cited document information acquisition unit 14 is a part of cited document information that is information indicating a document identified by the cited document identification information included in all or part of the document information acquired by the document information acquisition unit 13. Thus, the part corresponding to the item determined in advance according to the citation format in the citation information included in all or part of the document information acquired by the document information acquisition unit 13 is acquired. In addition, the cited document information acquisition unit 14 acquires a part corresponding to an item determined in advance according to the citation format. When the citation format is a predetermined format, a part of the cited document information is obtained. This includes cases where no acquisition is made.

  Here, the citation information is information indicating a description of a document identified by the cited document identification information. In the document information, for example, the citation information may exist in the same area as the cited document identification information, and the citation information may exist in a different area from the cited document identification information. In the former case, for example, cited document identification information is described in the background art area, and the cited document is cited by the cited document identification information itself. Such citations may be made in patent specifications and the like. On the other hand, in the latter case, for example, the cited document identification information is collectively described in the column of the reference, and in the text of the document, the predetermined reference corresponding to the cited document identification information described in the column of the reference is given. This is a case where the cited document is cited by the index. In academic literature, such citations are often made. Here, the index is, for example, 1, 2, 3 ..., 1), 2), 3) ..., [1], [2], [3] ..., [A], [B], [C ], Patent Document 1, Patent Document 2, Patent Document 3,..., Non-Patent Document 1, Non-Patent Document 2, Non-Patent Document 3,... It may be a thing. When a cited document is cited by the cited document identification information itself, the cited information is information including the cited document identification information. The information including the cited document identification information is, for example, information on a predetermined area including the cited document identification information. The predetermined area may be, for example, a sentence, a paragraph, or a group of other documents. When a cited document is cited by a predetermined index corresponding to the cited document identification information, the cited information is information including the index. The information including the index is, for example, information on a predetermined area including the index. The predetermined area may be, for example, a sentence, a paragraph, or a group of other documents. In the document information, when the citation information and the cited document identification information exist in different areas, and the document information acquisition unit 13 acquires a part of the document information, the document information acquisition unit 13 A part of the document information including a part of the document information may be acquired, or two or more parts of the document information including both may be acquired. Further, when the document information acquisition unit 13 acquires all or a part of the document information having the citation information, the citation information may be manifested in all or a part of the document information, or may not be manifested. May be. In the former case, the citation information part can be easily understood in all or part of the document information.

  Next, the citation format in the citation information will be described. The citation format includes, for example, a problem-pointing citation format (hereinafter also referred to as a “type C” citation format) that points out the problem of the cited document, and a format that takes the cited document as the basis of the editorial. There are certain editorial-based citation formats (hereinafter also referred to as “type B” citation formats) and other citation formats (hereinafter also referred to as “type O” citation formats). The type B citation format is a citation format used when a new theory is proposed or a system is constructed using existing research results. It should be noted that other citation formats may be used. The cited document information acquisition unit 14 holds, for example, one or more clue phrases corresponding to each citation format in a recording medium (not shown), and determines whether each of the citation information includes a clue phrase at a predetermined position. The citation format may be determined. The predetermined position may be, for example, the beginning of the citation information, the end of the citation information, or all of the citation information. Any method other than this may be used as long as the citation format can be determined from the citation information.

  Next, operation | movement of the related vocabulary acquisition apparatus 1 by this Embodiment is demonstrated using a flowchart. The overall operation of the related term acquisition apparatus 1 according to the present embodiment is the same as that of the flowchart of FIG. 3 is also the same as the flowchart of FIG. 4 of the first embodiment, and a description thereof will be omitted. However, it is assumed that the document information acquisition unit 13 also acquires citation information in step S204 of FIG. Also, the flowchart showing the details of the related term information acquisition process in step S104 in the flowchart of FIG. 3 is the same as the flowchart of FIG.

  FIG. 11 is a flowchart showing details of the cited document acquisition processing in step S103 in the flowchart of FIG. The processes other than steps S601 to S603 are the same as those in the flowchart of FIG. 5 of the first embodiment, and the description thereof is omitted.

  (Step S601) The cited document information acquisition unit 14 acquires all the cited document identification information included in the i-th document information acquired by the document information acquisition unit 13 and the cited information corresponding to the cited document identification information. . In addition, when the part of the citation information is not obvious in all or a part of the document information acquired by the document information acquisition unit 13, it is the same as the description in the citation information acquisition process by the document information acquisition unit 13. The cited document information acquisition unit 14 can specify the portion of the cited information.

  (Step S602) The cited document information acquisition unit 14 determines a citation format in the citation information corresponding to the j-th cited document identification information. Determining the citation format may be, for example, storing the j-th cited document identification information and information indicating the citation format in association with each other in a recording medium (not shown). Details of this processing will be described later with reference to the flowchart of FIG.

  (Step S603) The cited document information acquisition unit 14 searches the document information storage unit 11 for cited document information identified by the j-th cited document identification information acquired in step S601, and sets one of the retrieved cited document information. A portion corresponding to the citation format determined in step S602 is acquired from the document information storage unit 11.

  FIG. 12 is a flowchart showing details of the citation format determination process in step S602 in the flowchart of FIG. As described above, the cited document information acquisition unit 14 holds one or more clue phrases corresponding to each citation format in a recording medium (not shown).

(Step S701) The cited document information acquisition unit 14 sets a counter k to 1.
(Step S702) The cited document information acquisition unit 14 determines whether or not the cited information includes the k-th cue phrase of type C. If so, the process proceeds to step S703; otherwise, the process proceeds to step S704.

(Step S703) The cited document information acquisition unit 14 determines the citation format as type C. Then, the series of processing ends, and the process proceeds to step S603 in the flowchart of FIG.
(Step S704) The cited document information acquisition unit 14 increments the counter k by 1.

(Step S705) The cited document information acquisition unit 14 determines whether or not a k-th type C clue phrase exists. And when it exists, it returns to step S702, and when it does not exist, it progresses to step S706.
(Step S706) The cited document information acquisition unit 14 sets the counter k to 1.

  (Step S707) The cited document information acquisition unit 14 determines whether the cited information includes the k-th cue phrase of type B. If so, the process proceeds to step S708; otherwise, the process proceeds to step S709.

(Step S708) The cited document information acquisition unit 14 determines the citation format as type B. Then, the series of processing ends, and the process proceeds to step S603 in the flowchart of FIG.
(Step S709) The cited document information acquisition unit 14 increments the counter k by 1.

  (Step S710) The cited document information acquisition unit 14 determines whether or not a k-th type B clue phrase exists. And when it exists, it returns to step S707, and when it does not exist, it progresses to step S711.

  (Step S711) The cited document information acquisition unit 14 determines the citation format as type O. Then, the series of processing ends, and the process proceeds to step S603 in the flowchart of FIG.

  Next, operation | movement of the related vocabulary acquisition apparatus 1 by this Embodiment is demonstrated using a specific example. In this specific example, the document information storage unit 11 stores about 12,000 full-text papers (document information) centered on the natural language processing field in Postscript and PDF format. Of these, approximately 8000 are included in the ACL Anology provided by ACL, and the remaining approximately 4000 were collected from natural language processing researchers and natural language processing laboratory laboratories in Japan and overseas. And article data (document information) extracted from a collection of international conference proceedings (CD-ROM). As described above, the document information stored in the document information storage unit 11 in this specific example is one type of document information called an academic paper.

  Further, in this specific example, the document information acquisition unit 13 corresponds to the information in the column of the reference document of the document information from the document information in which the term indicated by the term information received by the term information receiving unit 12 is included in the title. Citation information shall be obtained.

  In this specific example, the cited document information acquisition unit 14 is a type C clue phrase dictionary that is information having a clue phrase corresponding to type C, and type B that is information having a clue phrase corresponding to type B. The clue phrase dictionary is held in a recording medium (not shown). FIG. 13 is a diagram illustrating an example of a type C clue phrase dictionary. FIG. 14 is a diagram illustrating an example of a type B clue phrase dictionary. The cited document information acquisition unit 14 determines that the citation format is type C when any of the clue phrases in the type C clue phrase dictionary is included in the citation information, and any of the type B clue phrase dictionaries is included in the citation information. If the clue phrase is included, it is determined that the citation format is type B.

In this specific example, the cited document information acquisition unit 14 acquires the title of the cited document as a part of the cited document.
First, it is assumed that the user inputs the term information “terminology” to the related term acquisition device 1. Then, the term information “terminology” is accepted by the term information accepting unit 12 (step S101). Then, the document information acquisition unit 13 acquires a reference field (Reference) column portion of each document information including the term information “terminology” in the title and citation information (steps S102 and S201 to S205).

  Specifically, the document information acquisition unit 13 searches each document information including the term information “terminology” in the title, and acquires the bibliographic information of the searched document information (step S201). FIG. 15 is a diagram illustrating an example of the bibliographic information of the acquired document information. As shown in FIG. 15, the bibliographic information includes “author”, “title”, and “other”. “Author” is information on the author (one or more) of the paper, “Title” is information on the title of the paper, and “Other” is information such as the name and year of the journal. In FIG. 15, “ID” is associated with each record. This “ID” exists for record management in the table.

  The document information acquisition unit 13 obtains from the document information identified by each record shown in FIG. 15 the citation information corresponding to the reference document (Reference) column part and the cited document identification information included in the reference document column. Are acquired (step S204).

  Here, a method for obtaining citation information in this specific example will be described. As described above, the document information acquisition unit 13 acquires a predetermined index corresponding to the cited document identification information described in the reference literature column from the reference literature column, and searches for a sentence including the index. Next, the document information acquisition unit 13 extracts a sentence closely related to the sentence including the index using the clue phrases (1) to (6) below. In other words, the document information acquisition unit 13, when any of the following clue phrases (1) to (6) is included in the sentence following the sentence including the index, the sentence including the clue phrase includes the index. Extracted as a sentence closely related to the sentence it contains. The document information acquisition unit 13 acquires a sentence including an index and a sentence closely related to the sentence including the index as citation information.

(1) Clues on anaphor: In this, On this, Such
(2) Clues related to conjunctions: But, However, Although
(3) First person clues: We, we, Our, our, us, I
(4) A clue about the third person: They, thee, Their, their, them
(5) Clues about adverbs: Furthermore, Additionally, Still
(6) Other clues: In particular, follow, For example

  Next, the cited document information acquisition unit 14 determines the citation format based on a part of the document information acquired by the document information acquisition unit 13 (that is, the citation information corresponding to the reference document part and each cited document identification information). The part of the cited document corresponding to the citation format is acquired (step S103, flowcharts of FIGS. 11 and 12).

  In the determination of the citation format, as described above, when any of the clue phrases in the type C clue phrase dictionary shown in FIG. 13 is included in the citation information, the citation format is determined as type C, and FIG. When any cue phrase in the type B cue phrase dictionary shown is included in the citation information, the citation format is determined as type B, and if it is neither type C nor type B, the citation format is typed. O is determined (steps S701 to S711). Then, when the citation format is type C, the cited document information acquisition unit 14 acquires information indicating the title of the cited document information as a part of the cited document information. On the other hand, when the citation format is type B, the cited document information acquisition unit 14 acquires information indicating a summary (abstract) of the cited document information as a part of the cited document information. The cited document information acquisition unit 14 does not acquire a part of the cited document information when the citation format is type O.

  Thereafter, the related term information output unit 16 acquires related term information from a partial set of cited reference information acquired by the cited reference information acquisition unit 14 (step S104, flowchart of FIG. 6), and outputs related term information. The unit 16 outputs the acquired related term information (step S105).

  Needless to say, the method for obtaining the citation information is not limited to the specific description described above. Here, another method for acquiring citation information will be described. For example, as described above, the document information acquisition unit 13 acquires a predetermined index corresponding to the cited document identification information described in the reference column from the reference column, and searches for a sentence including the index. Identify the sentence that contains the index. Then, the document information acquisition unit 13 acquires a sentence including the index and a sentence closely related to the sentence as citation information. Here, a method for identifying a sentence closely related to the sentence including the index will be described.

It is assumed that the document information acquisition unit 13 can access the following two types of clue words and phrases stored in advance on a recording medium (not shown).
(A) Pre-extended cue phrases In this, On this, Such, But, However ...
(B) Cue phrases for post-expansion We, we, They, they, Our, our, Inthis, Onthis, Such, Bute, However ...

  First, the document information acquisition unit 13 sets a sentence including an index as a candidate sentence for citation information. Then, when at least one of the preceding extension clue words / phrases is included in the first sentence of the candidate sentence, the sentence one sentence before the candidate sentence is included in the candidate sentence. The literature information acquisition unit 13 repeats the process of extending the candidate sentence before the first sentence of the candidate sentence does not include all the (A) pre-extended clue words. In addition, the document information acquisition unit 13 includes (B) a post-extension clue phrase in a sentence next to the candidate sentence (a sentence adjacent to the candidate sentence and not included in the candidate sentence). In the case, the sentence following the candidate sentence is included in the candidate sentence. The document information acquisition unit 13 repeats the process of expanding the candidate sentence later until the next sentence after the candidate sentence does not include all of the post-expansion clue words (B). The candidate sentence at the time when the process of extending the candidate sentence is completed becomes the citation information. The literature information acquisition unit 13 acquires the citation information. Note that (A) the pre-extension cue word / phrase and (B) the post-extension cue word / phrase are examples, and if the sentence closely related to the sentence including the index can be appropriately extracted, A clue word may be used. In addition, here, attention is focused on the sentence including the index. However, in the technical document that directly refers to the cited document by the cited document identification information, the sentence including the cited document identification information is changed to the sentence including the index. It will be used instead.

  Needless to say, the method of determining the citation format is not limited to the specific description described above. Here, another method for determining the citation format will be described. For example, the document information acquisition unit 13 determines the citation format in consideration of the position of the cue phrase included in the citation information, not only based on whether or not the citation information includes a predetermined clue phrase. May be. More specifically, for example, in the type C cue phrase dictionary, as shown in FIG. 20, the type C cue phrase (in FIG. 20, “cue phrase”) corresponds to the position of the cue phrase. It shall be attached. The “position of the clue phrase” in FIG. 20 is information indicating the number of sentences in the citation information with the sentence including the index and the cited document identification information as a reference (0). For example, in the case of the position “1 to 3” of the clue phrase, the index or the cited document identification information is started from the sentence (position of the clue phrase = 1) after the sentence including the index or the cited document identification information in the citation information. Means three sentences from the sentence including the first sentence to the third sentence (cue phrase position = 3). When the position of the clue phrase is negative, it means a sentence before the sentence including the index and the cited document identification information. When the type C clue phrase is included in the sentence indicated by the position of the corresponding clue phrase, the document information acquisition unit 13 may determine that the citation format of the citation information is type C. More specifically, the document information acquisition unit 13 adds “in the citation information to any sentence from the sentence (first sentence) to the third sentence after the sentence including the index and the cited document identification information”. When “However” is included, the citation format of the citation information may be determined to be type C. Here, the determination method for determining whether or not the citation format is type C has been described, but the determination method for determining whether or not the citation format is type B is the same.

  Thus, according to the related vocabulary acquisition apparatus 1 according to the present embodiment, a part of cited document information corresponding to an item corresponding to a citation format of a cited document can be acquired. By appropriately setting whether to obtain a part of cited reference information corresponding to an item, it is possible to obtain related term information with higher accuracy.

In the present embodiment, as in the above specific example, when the citation format is type C, information indicating the title of the cited document information is acquired, and when the citation format is type B, Although the case where information indicating the summary of cited document information is acquired has been described, this may be another setting. For example, when the citation format in the citation information included in all or part of the document information acquired by the document information acquisition unit 13 is the problem-pointing citation format, the cited document information acquisition unit 14 When the citation format in the citation information included in all or part of the document information acquired by the document information acquisition unit 13 is an editorial basis citation format, A portion having a typical item may be acquired. Here, the background item may be, for example, “prior art” or “background technology” if it is a patent document, and “introduction” or “related research” if it is an academic document. There may be. The characteristic item may be, for example, “embodiment” or “effect” in the case of patent documents, and “at the end” or “conclusion” in the case of academic literature. May be. The method described in the first and second embodiments can be used as a method of obtaining the cited document information portion corresponding to these items.
Further, it goes without saying that the process of acquiring the portion of the cited document information corresponding to the citation format may be executed in the related term acquisition device 2 in the second embodiment.

  Further, in each of the above embodiments, the document information stored in the document information storage unit 11 may be two or more types. For example, in the document information storage unit 11, at least a first type of document information and a second type of document information are stored, and the document information acquisition unit 13 is stored in the document information storage unit 11. All or part of the first type of document information is acquired from the first type of document information, and the cited document information acquisition unit 14 stores the second type of document information stored in the document information storage unit 11. From the above, a part of the cited document information may be acquired. For example, the first type of document information may be information relating to academic papers, and the second type of document information may be information on patent documents, or vice versa. In general, users who are accustomed to academic papers are often not familiar with the terms used in patent documents. However, by using such a related term acquisition device, the terms used in academic papers can be related to those terms. Related terms used in patent documents. Further, the first type of document information may be information of the first type of language, and the second type of document information may be information of the second type of language. The first type language and the second type language may be any language such as English, Japanese, Chinese, French, German, and the like. In this way, for example, it may be possible to acquire English related terms related to Japanese terms.

(Embodiment 4)
A technical trend information creating apparatus according to Embodiment 4 of the present invention will be described with reference to the drawings.
FIG. 16 is a block diagram showing the configuration of the technical trend information creation device according to this embodiment. In FIG. 16, the technical trend information creation device 1001 according to the present embodiment includes a search term receiving unit 1011, a document information storage unit 1012, a search unit 1013, a specifying unit 1014, a document information acquisition unit 1015, and technical terms. An extraction unit 1016, a time information acquisition unit 1017, a technology trend information creation unit 1018, and an output unit 1019 are provided.

  The search term receiving unit 1011 receives a search term. A search term is information used as a search keyword in a search in the search unit 1013 described later. The search term accepted by the search term acceptance unit 1011 is actually information indicating the search term, but is simply referred to as a search term. In the technical trend information creation device 1001, the technical literature is searched based on the search term received by the search term receiving unit 1011. Information indicating the searched technical literature is acquired, and the technical trend information is obtained based on the information. Will be created. As described above, the search term is used to determine a field or the like in which the technical trend information is created. This search term is generally text data, but may be other information. Further, this search term may be, for example, information indicating a field of created technical trend information, or information indicating a period of the generated technical trend information. When the search term is information indicating a technical trend information field, the search term may be, for example, the name of the field or a symbol associated with the field. Examples of symbols associated with the fields include IPC (international patent classification) and US class (US patent classification). In the present embodiment, a case will be described in which the search term receiving unit 1011 receives a search term indicating the name of the field of technology trend information.

  Here, the search term acceptance by the search term acceptance unit 1011 may be, for example, acceptance of information input from an input device (for example, a keyboard, a mouse, a touch panel, etc.), which is transmitted via a wired or wireless communication line. May be received, or information read from a predetermined recording medium (for example, an optical disk, a magnetic disk, a semiconductor memory, etc.) may be received. Note that the search term receiving unit 1011 may or may not include a device (for example, a modem or a network card) for receiving. In addition, the search term receiving unit 1011 may be realized by hardware, or may be realized by software such as a driver that drives a predetermined device.

  The document information storage unit 1012 stores at least a plurality of document information and a plurality of time information. Here, the literature information is information indicating at least a part of the technical literature. The technical literature includes, for example, technical papers, patent / utility model publications, published technical bulletins, technical-related books and articles, technical web pages, and the like. One document information includes at least a part of information (data) of the corresponding one technical document. The part of the technical document is a part of the technical document that includes at least a range necessary for subsequent processing. For example, when the title of a technical document and a reference document are used in a later process, the document information may be information indicating the title of the technical document and the reference document. Note that the document information may be information indicating all technical documents. Moreover, time information is information which shows the time regarding the technical literature which literature information shows. The time information may be, for example, information indicating the time when the document information was created, information indicating the time when the document information was released, or information indicating other time related to the document information. Also good. The time information may be, for example, information indicating a year regarding document information, information indicating a date, information indicating a date, or information indicating another time. It is assumed that the document information and the time information are associated directly or indirectly so that it can be understood which technical document is at which time. In the former case, for example, document information and time information corresponding to each other may be included in the same record. In the latter case, for example, the document information and the time information may be associated with each other via a document ID for identifying a technical document. The document information and the time information may be a text file, or may be a file of a unique format such as word processing software. The document information and the time information may be integrated as data, or may be separate data.

  It goes without saying that other information may be stored in the document information storage unit 1012. For example, bibliographic information such as author name, publisher name, medium name (eg, magazine name, book name, etc.), technical literature field (eg, IPC), technical literature summary, etc. It may be stored in the storage unit 1012. The document information may include, for example, cited document identification information or citation information. The cited document identification information is information for identifying another technical document cited in the document information (hereinafter, this technical document may be referred to as “cited document”). The citation information is information indicating a description of another technical document cited in the document information including the citation information. The description about a technical document is, for example, a description for pointing out a problem of the technical document, a description for proposing a new theory based on the technical document, or constructing a system. Here, the content of the cited document identification information is not limited as long as it is information that can identify the cited document. The cited document identification information may be, for example, an ID for identifying a cited document such as a patent number, an application publication number of a patent application, or a PMID (PubMed Identifier), and relates to an author name, title, date of writing, and medium It may be bibliographic information of a document that can identify a cited document such as information (magazine name, publisher name, URL, etc.). In the present embodiment, it is assumed that the document information includes at least a title, cited document identification information, and cited information.

  Here, the cited document identification information and the cited information will be described in more detail. In the document information, for example, the citation information may exist in the same area as the cited document identification information, and the citation information may exist in a different area from the cited document identification information. In the former case, for example, cited document identification information is described in the background art area, and the cited document is cited by the cited document identification information itself. Such citations may be made in patent specifications and the like. On the other hand, in the latter case, for example, the cited document identification information is collectively described in the column of the reference, and in the text of the document, the predetermined reference corresponding to the cited document identification information described in the column of the reference is given. This is a case where the cited document is cited by the index. In academic literature, such citations are often made. Here, the index is, for example, 1, 2, 3 ..., 1), 2), 3) ..., [1], [2], [3] ..., [A], [B], [C ], Patent Document 1, Patent Document 2, Patent Document 3,..., Non-Patent Document 1, Non-Patent Document 2, Non-Patent Document 3,... It may be a thing. When a cited document is cited by the cited document identification information itself, the cited information is information including the cited document identification information. The information including the cited document identification information is, for example, information on a predetermined area including the cited document identification information. The predetermined area may be, for example, a sentence, a paragraph, or a group of other documents. When a cited document is cited by an index corresponding to the cited document identification information, the cited information is information including the index. The information including the index is, for example, information on a predetermined area including the index. The predetermined area may be, for example, a sentence, a paragraph, or a group of other documents. This predetermined area may be determined based on a predetermined rule, as will be described later.

  Next, the citation format of the cited document in the citation information will be described. The citation format includes, for example, a problem-pointing citation format (hereinafter also referred to as a “type C” citation format) that points out the problem of the cited document, and a format that takes the cited document as the basis of the editorial. There are certain editorial-based citation formats (hereinafter also referred to as “type B” citation formats) and other citation formats (hereinafter also referred to as “type O” citation formats). The type B citation format is a citation format used when a new theory is proposed or a system is constructed using existing research results. It should be noted that other citation formats may be used. Here, the citation form of the above description is described in the following document.

  References: Eiwa Namba, Manabu Okumura, “Development of Survey Paper Creation Support System Considering Reference Information Between Papers”, Natural Language Processing, Vol. 6, no. 5, p. 43-62, 1999

  The document information storage unit 1012 can be realized by a predetermined recording medium (for example, a semiconductor memory, a magnetic disk, an optical disk, etc.). The process of storing document information or the like in the document information storage unit 1012 is not limited. For example, document information or the like may be stored in the document information storage unit 1012 via a recording medium, and document information or the like transmitted via a communication line or the like may be stored in the document information storage unit 1012. Alternatively, the document information or the like input via the input device may be stored in the document information storage unit 1012.

  The search unit 1013 searches the document information storage unit 1012 using the search term received by the search term receiving unit 1011. The search unit 1013 preferably searches for document information corresponding to a technical document whose search term received by the search term receiving unit 1011 is the main topic. “Literature information corresponding to technical literature” means literature information indicating at least a part of the technical literature. The main topic is the name of a research field. The extent of the range can vary depending on the range of interest. For example, “morphological analysis” can be a main topic, but “natural language processing” including it can also be a main topic.

  For such a search, the search unit 1013 may search the entire document information stored in the document information storage unit 1012, or search for a part of each document information (for example, only the title). Alternatively, information other than literature information may be searched. When searching a part of the document information, the document information has a structure capable of specifying a part of the document information. For example, the item name may be described for each part of the document information, or the item may be specified using a markup language tag or the like. “Information other than document information” may be, for example, time information or other bibliographic information.

  Further, in order to search for document information corresponding to the technical document whose search term is the main topic, the search unit 1013 may search for document information in which the search term is described more than a predetermined number of times, Or you may search the literature information in which the search term is described in the predetermined several area | region (for example, a title, summary, etc.). As described above, a method for searching for document information corresponding to a technical document in which a certain search term is a main topic is already known in the following documents and the like, and detailed description thereof will be omitted.

  Literature: Richard Baeza-Yates, Bertier Ribeiro-Neto, “Modern Information Retrieval”, Addison-Wesley Pub, ISBN: 0201339829X, 1999.

  Literature: Takenobu Tokunaga, “Information Retrieval and Language Processing”, University of Tokyo Press, ISBN: 4130654055, 1999

  Literature: Nitin Indurkhya, Tong Zhang, Fred J. Damerau, “Text Mining: Predictive Methods for Analyzing Information”, Springer, ISBN: 0387954333, 2004.

  When the document information is information including at least the title of the technical document, the search unit 1013 searches the area indicating the main topic in the title of the technical document indicated by the document information stored in the document information storage unit 1012. The document information including the search term received by the term receiving unit 1011 may be searched. By doing in this way, it is thought that the literature information corresponding to the technical literature whose search term is the main topic can be searched. The area indicating the main topic in the title may be the last area of the title in a Japanese title, for example, and may be the first area of the title in an English title.

  Note that the search processing itself performed by the search unit 1013 is already known, and thus detailed description thereof is omitted. As a result of the search by the search unit 1013, generally one or more document information is specified as the search result. The identification may be performed, for example, by accumulating information identifying the document information in a recording medium such as a temporary storage memory, or by setting a flag in association with the document information. .

  The specifying unit 1014 specifies the document information having the same field as the document information searched by the search unit 1013. By specifying by the specifying unit 1014, the document information that the search unit 1013 has not searched for is also used to create the technical trend information. Will be able to. Here, as the identification method, there are, for example, (1) a method using citation relationships, (2) a method using document classification and document clustering, and the like.

(1) Method Using Citation Relationship In the method using citation relationship, the document information corresponding to the technical document cited in the document information searched by the search unit 1013 has the same field as the document information searched by the search unit 1013. This is a method of identifying document information on the assumption that there is. In specifying the document information, there are, for example, (1-1) a method using only cited document identification information and (1-2) a method using cited document identification information and a citation format.

(1-1) Method using only cited document identification information The identifying unit 1014 is a document identification that is information for identifying other technical documents cited in the document information from the document information retrieved by the retrieval unit 1013. Get information. Then, the document information is specified by the acquired cited document identification information.

  In the acquisition of the cited document identification information, the specifying unit 1014 may acquire a column of a reference document (Reference) that is considered to include the cited document identification information, or may acquire a column of patent documents and non-patent documents. Alternatively, a region where the cited document identification information exists may be acquired by using a technique such as pattern matching. When acquiring a reference column, patent document, or the like, for example, information indicating an item corresponding to a part to be acquired is stored in a recording medium (not shown), and the specifying unit 1014 extracts an item from the recording medium. Information indicated may be read out, and an area of literature information corresponding to the item may be acquired. Further, when the cited document identification information is acquired using a technique such as pattern matching, for example, information indicating a pattern used for pattern matching or the like is stored in a recording medium (not shown), and the specifying unit 1014 An area including the cited document identification information in the document information may be acquired by reading a pattern or the like from the recording medium and performing pattern matching or the like. The cited document identification information acquired by the specifying unit 1014 may be temporarily stored in a recording medium (not shown).

(1-2) Method of Using Cited Document Identification Information and Citation Format The specifying unit 1014 acquires the cited document identification information from the document information searched by the search unit 1013, and is cited in the document information. Citation information, which is information indicating a description of the technical document, is also acquired. If the citation format indicated by the obtained citation information is a problem indication type citation format (type C citation format) that indicates the problem of the cited technical document, it is described in the citation information. The document information is specified by the cited document identification information for identifying the technical document being read.

  For example, when the technical document B is cited in a technical document A, the technical document A is cited when the technical document B is cited in a format (type C citation format) that points out the problem of the technical document B. And the technical literature B are found to be likely to have the same field (see the above-mentioned literature “Development of Survey Paper Creation Support System Considering Reference Information Between Papers”).

  Therefore, when the citation format of the technical document indicated by the citation information is a type C citation format, the document information is specified by the cited document identification information for identifying the technical document, and if not, the document information is specified. It does not have to be.

  The specifying unit 1014 holds, for example, one or more clue words / phrases according to the type C citation format in a recording medium (not shown), and the citation format depends on whether the cued word / phrase is included in a predetermined position of the citation information. May be determined to be type C. The predetermined position may be, for example, the beginning of the citation information, the end of the citation information, or all of the citation information, or a sentence ( For example, a predetermined range may be used based on a sentence including cited document identification information or a sentence including an index corresponding to the cited document identification information. The predetermined position may be set for each clue word. Note that other methods may be used as long as the citation format can be determined from the citation information.

  In the acquisition of the citation information, the specifying unit 1014 may specify, for example, an area referring to the cited document identification information using the cited document identification information, and acquire the citation information including the specified area. Alternatively, an area referring to the index corresponding to the cited document identification information may be specified using the index corresponding to the cited document identification information, and the citation information including the specified area may be acquired. The citation information acquired by the specifying unit 1014 may be temporarily stored in a recording medium (not shown).

  When the identifying unit 1014 acquires the citation information and determines that the citation format in the citation information is type C, the identifying unit 1014 identifies the document information with the cited document identification information that identifies the technical document described in the citation information. Will do. For example, in the cited information, when the cited document is referred to by the cited document identification information, the document information may be specified by the cited document identification information. On the other hand, in the cited information, when the cited document is referred to by the index, the cited document identification information corresponding to the index is identified, and the document information is identified by the identified cited document identification information. .

  In the acquisition of the cited document identification information and the cited information in the above (1-1) and (1-2), the specifying unit 1014 may acquire the cited document identification information or the like from the document information storage unit 1012. When the document information acquisition unit 1015 described later acquires document information including cited document identification information and the like, the citation information may be acquired from the document information acquired by the document information acquisition unit 1015. In the present embodiment, the former case will be described.

(2) Method Using Document Classification or Document Clustering A method for classifying or clustering a plurality of document information has been conventionally known. For example, it is disclosed in the above-mentioned documents “Modern Information Retrieval”, “Information Retrieval and Language Processing”, “Text Mining: Predictive Methods for Analyzing Information”, and the like. In this method, the document information classified into the same group or clustered is identified on the assumption that the document information is the same field. More specifically, the document information classified into the same group as the document information searched by the search unit 1013 or clustered is specified, and the specifying unit 1014 has the same field as the document information searched by the search unit 1013. The literature information of is specified. The identification of the document information in this case may be performed by, for example, a document ID for identifying the document information or bibliographic information of the document information (title, publication year, first author name, medium name on which the document is published, etc.).

  The document information acquisition unit 1015 acquires all or a part of at least a part of the document information among the plurality of document information stored in the document information acquisition unit 1015. The document information acquisition unit 1015 acquires all or part of the document information searched by the search unit 1013. When the document information acquisition unit 1015 acquires part of the document information, the part of the document information may be, for example, a title or summary of the document information. For example, information indicating an item corresponding to the part to be acquired is stored in a recording medium (not shown), and the document information acquiring unit 1015 reads information indicating the item from the recording medium, and the document information corresponding to the item May be acquired as part of the document information. When the document information acquisition unit 1015 acquires a part of the document information and the search unit 1013 searches for a part of the document information, a part of the document information to be acquired and a part of the document information used in the search The same part of the document information may be used, or a different part of the document information may be used.

The document information acquisition unit 1015 also acquires all or part of the document information specified by the specifying unit 1014 from the document information storage unit 1012. For example, when the document information is specified by the cited document identification information, the document information acquisition unit 1015 searches the document information storage unit 1012 using the cited document identification information, and all of the retrieved document information. Or get some. Here, the document information area acquired by the document information acquisition unit 1015 using the specified result by the specifying unit 1014 is the same as the document information area acquired by the document information acquiring unit 1015 using the search result by the search unit 1013. Suppose there is. For example, when the document information acquisition unit 1015 acquires the title of the document information using the search result, the document information acquisition unit 1015 acquires the title of the specified document information.
The document information acquisition unit 1015 may temporarily store all or part of the document information acquired according to the search result or the specific result in a recording medium (not shown).

  The technical term extraction unit 1016 extracts technical terms from all or part of the document information acquired by the document information acquisition unit 1015. The document information acquired by the document information acquisition unit 1015 is the document information acquired by the document information acquisition unit 1015 partially or entirely. Here, the technical term is a term used in the technical literature field searched by the search unit 1013. This technical term is preferably a technical term that is an element in the technical literature field searched by the search unit 1013. By looking at the transition of this technical term, it is possible to know the technical trend. The technical term extracted by the technical term extraction unit 1016 is actually information indicating the technical term, but is simply referred to as a technical term. A method by which the technical term extraction unit 1016 extracts technical terms will be described later. The technical term extraction unit 1016 may temporarily store the extracted technical terms in a recording medium (not shown).

  The time information acquisition unit 1017 acquires time information corresponding to the document information including the technical term extracted by the technical term extraction unit 1016 from the document information acquisition unit 1015. For example, when the time information acquisition unit 1017 acquires time information corresponding to the document information acquired by the document information acquisition unit 1015, the time information acquisition unit 1017 can acquire the above-described time information. In this case, when a technical term is not extracted from certain document information, it is not necessary to acquire time information corresponding to the document information, or a technical trend that describes time information corresponding to the document information later. It may not be used in the creation of information.

  The technical trend information creation unit 1018 creates technical trend information using the technical terms extracted by the technical term extraction unit 1016 and the timing information acquired by the timing information acquisition unit 1017. Here, the technical trend information is information that associates a technical term with a period related to a technical document including the technical term. That is, it is possible to know when a certain technical term was used by the technology trend information. In the technical trend information, the time associated with the technical term may be the time information itself (for example, the numerical value itself such as the year or month), or different from the time information (for example, the year, month, etc.) Graphic etc.). In the latter case, for example, a time associated with the technical information may be indicated using a coordinate axis indicating the time. The technical trend information may be in any data format as long as the technical term extracted by the technical term extraction unit 1016 and the time information acquired by the time information acquisition unit 1017 are associated with each other. For example, it may be a text file, a spreadsheet format file, an image file, or another format file. This technical trend information may be, for example, an image indicating the time indicated by the time information corresponding to the document information including the technical term for each technical term. The technical trend information may include a search term accepted by the search term acceptance unit 1011.

  The output unit 1019 outputs the technology trend information created by the technology trend information creation unit 1018. Here, the output may be, for example, display on a display device (for example, a CRT or a liquid crystal display), transmission via a communication line to a predetermined device, printing by a printer, or output to a recording medium. It may be accumulated. The output unit 1019 may or may not include an output device (for example, a display device or a printer). The output unit 1019 may be realized by hardware, or may be realized by software such as a driver that drives these devices. For example, when the technology trend information is an image, the output unit 1019 displays and outputs the technology trend information.

  In addition, the language used in this technical trend information creation apparatus 1001 is not ask | required. For example, the language may be Japanese, English, French, Chinese, Korean.

Next, the operation of the technical trend information creation apparatus 1001 according to this embodiment will be described using the flowchart of FIG.
(Step S1101) The search term receiving unit 1011 determines whether a search term has been received. If a search term is accepted, the process proceeds to step S1102, and if not, the process of step S1101 is repeated until a search term is accepted.

  (Step S1102) The search unit 1013 searches the document information storage unit 1012 with the search terms received by the search term receiving unit 1011. For example, the search unit 1013 may acquire a document ID that identifies the searched document information.

  (Step S1103) The identifying unit 1014 identifies document information having the same field as the document information searched by the search unit 1013. Details of this process will be described later.

  (Step S1104) The document information acquisition unit 1015 acquires all or part of the document information searched by the search unit 1013 and all or part of the document information specified by the specifying unit 1014 from the document information storage unit 1012. Details of this process will be described later.

  (Step S1105) The technical term extraction unit 1016 extracts technical terms from all or a part of the document information acquired by the document information acquisition unit 1015. Details of this processing will be described later.

  (Step S1106) The time information acquisition unit 1017 acquires the time information corresponding to the document information including the technical term extracted by the technical term extraction unit 1016 from the document information storage unit 1012.

  (Step S1107) The technical trend information creation unit 1018 creates technical trend information using the technical terms extracted by the technical term extraction unit 1016 and the timing information acquired by the timing information acquisition unit 1017.

(Step S1108) The output unit 1019 outputs the technology trend information created by the technology trend information creation unit 1018. Then, the process returns to step S1101. If the technology trend information can be output continuously as in the case of display output, the output may be terminated after a certain period of time has passed, or the termination instruction, interrupt processing, etc. The output may be terminated with the reception of the message as a trigger.
In the flowchart of FIG. 17, the process ends when the power is turned off or the process ends.

  Also, nothing is searched in the search process (step S1102), the document information acquisition process (step S1104), the technical term extraction process (step S1105), the time information acquisition process (step S1106), etc. in the flowchart of FIG. If no information is obtained or no document information is acquired, it may be determined that an error has occurred, and the process may return to step S1101.

  Even when the document information is not specified in the document information specifying process (step S1103), the process may proceed to step S1104 or may return to step S1101. In the present embodiment, the former case will be described.

  Further, the order of processing in the flowchart of FIG. 17 is arbitrary to some extent. For example, the process for acquiring the time information may be performed before the process for acquiring the document information or the process for extracting the technical terms as long as it is after the process for specifying the document information.

Next, detailed processing of the document information identification processing (step S1103) in the flowchart of FIG. 17 will be described with reference to the flowchart of FIG. As described above, there are various methods for specifying document information. Here, a method for specifying document information using cited document identification information and a cited format will be described.
(Step S1201) The identifying unit 1014 sets the counter i to 1.

  (Step S1202) The identifying unit 1014 acquires all the cited document identification information and the cited information corresponding to each cited document identification information from the i-th document information searched by the search unit 1013. As described above, the cited document identification information may or may not be included in the cited information. Further, in this step, when the cited document identification information is not included at all in the i-th document information searched by the search unit 1013, the process proceeds to step S1203 without acquiring the cited document identification information.

(Step S1203) The identifying unit 1014 sets the counter j to 1.
(Step S1204) The identifying unit 1014 determines whether or not the j-th cited document identification information acquired from the searched i-th document information exists. If it exists, the process proceeds to step S1205; otherwise, the process proceeds to step S1208.

  (Step S1205) The identifying unit 1014 determines whether the citation format of the citation information corresponding to the j-th cited document identification information is type C. If it is type C, the process proceeds to step S1206. If not, the process proceeds to step S1207.

  (Step S1206) The specifying unit 1014 temporarily stores the j-th cited document identification information in a recording medium (not shown). The document information identified by the temporarily stored cited document identification information becomes the identified document information.

(Step S1207) The identifying unit 1014 increments the counter j by 1. Then, the process returns to step S1204.
(Step S1208) The identifying unit 1014 increments the counter i by 1.

  (Step S1209) The specifying unit 1014 determines whether or not the i-th document information searched by the search unit 1013 exists. If it exists, the process returns to step S1202, and if not, the series of processes for specifying the document information ends, and the process proceeds to step S1104.

Next, detailed processing of the document information acquisition processing (step S1104) in the flowchart of FIG. 17 will be described using the flowchart of FIG.
(Step S1301) The document information acquisition unit 1015 acquires all or part of each document information indicated by the search result by the search unit 1013. Details of this processing will be described later.

  (Step S1302) The document information acquisition unit 1015 determines whether or not the document information is specified by the specifying unit 1014. If the document information is specified, the process proceeds to step S1303. If not, the series of processes for obtaining the document information ends, and the process proceeds to step S1105.

  (Step S1303) The document information acquisition unit 1015 acquires all or part of the document information specified by the specifying unit 1014. Details of this process will be described later. When this process ends, a series of processes for acquiring document information ends, and the process proceeds to step S1105.

  Next, processing for acquiring all or part of each piece of document information indicated by the search result in the flowchart of FIG. 19 will be described with reference to the flowchart of FIG.

(Step S1401) The literature information acquisition unit 1015 sets the counter i to 1.
(Step S1402) The document information acquisition unit 1015 acquires all or part of the document information indicated by the i-th search result.

(Step S1403) The document information acquisition unit 1015 temporarily stores all or part of the document information acquired in Step S1402 in a recording medium (not shown). This recording medium may be included in the document information acquisition unit 1015 or may be included in other components.
(Step S1404) The document information acquisition unit 1015 increments the counter i by 1.

  (Step S1405) The document information acquisition unit 1015 determines whether or not the i-th search result exists. If it exists, the process returns to step S1402. If not, the series of processes for acquiring all or part of each document information indicated by the search result ends, and the process proceeds to step S1302.

Next, processing for acquiring all or part of the specified document information in the flowchart of FIG. 18 will be described with reference to the flowchart of FIG.
(Step S1501) The literature information acquisition unit 1015 sets the counter i to 1.

  (Step S1502) The document information acquisition unit 1015 determines whether all or a part of the i-th document information specified by the specifying unit 1014 has not been acquired yet. If not acquired, the process proceeds to step S1503. If acquired, the process proceeds to step S1505. This determination can be made, for example, by referring to a recording medium (not shown) in which the document information acquired by the document information acquisition unit 1015 in step S1403 is temporarily stored.

(Step S1503) The document information acquisition unit 1015 acquires all or part of the i-th document information specified by the specifying unit 1014.
(Step S1504) The document information acquisition unit 1015 temporarily stores all or part of the document information acquired in Step S1503 in a recording medium (not shown). This recording medium may be included in the document information acquisition unit 1015 or may be included in other components.

(Step S1505) The literature information acquisition unit 1015 increments the counter i by 1.
(Step S1506) The document information acquisition unit 1015 determines whether or not the i-th document information specified by the specifying unit 1014 exists. If it exists, the process returns to step S1502, and if not, the series of processes for acquiring the specified document information ends, and the process proceeds to step S1105.

Next, a method for extracting technical terms from all or part of the document information acquired by the document information acquisition unit 1015 will be described.
First, the basic principle of extracting technical terms will be described. It is assumed that all or a part of the document information acquired by the document information acquisition unit 1015 includes technical terms in the field of the main topic that has been searched or specified. The technical terms are generally considered to often correspond to predetermined clue phrases. In particular, when the name of the main topic is also included in all or part of the document information acquired by the document information acquisition unit 1015, the technical term is associated with the name of the main topic via a predetermined clue phrase. It is thought that there are many cases.

  For example, when the name of the main topic is “sentence extraction” and the document information acquisition unit 1015 acquires part of the document information “sentence extraction using SVM”, the name of the main topic “sentence extraction” The technical term “SVM” in that field is associated with the clue word “using”. Further, for example, when the name of the main topic is “caption generation” and the document information acquisition unit 1015 acquires a part of the document information “sentence reduction for caption generation”, the name of the main topic “caption” It can be seen that the “generation” is associated with the technical term “sentence shortening” in the field via the clue word “for”. In addition, even when the name of the main topic is not included in all or part of the document information acquired by the document information acquisition unit 1015, it is possible to specify the technical term by using the clue word / phrase. Conceivable.

  Therefore, identifying those clue words, searching all or part of the document information acquired by the document information acquisition unit 1015 using the clue words, and extracting the terms corresponding to the searched clue words as technical terms Can be done. In addition, if it is only that method, there is a possibility that an inappropriate technical term may be extracted, so the inventors have also considered a measure for increasing the accuracy. Hereinafter, the configuration and operation of the technical term extraction unit 1016 will be specifically described.

[Pattern 1]
FIG. 22 is a block diagram illustrating a configuration of the technical term extraction unit 1016 in the pattern 1. In FIG. 22, the technical term extraction unit 1016 includes a clue phrase storage unit 1021, a clue phrase search unit 1022, and a technical term extraction unit 1023. Note that all or part of the document information acquired by the document information acquisition unit 1015 is stored in the document information temporary storage unit 1020. For example, the document information temporary storage unit 1020 may be included in the document information acquisition unit 1015, the technical term extraction unit 1016, or may be included in other components.

  The clue phrase storage unit 1021 stores one or more clue phrases. As described above, this clue word / phrase is a clue word / phrase to be used for extracting technical terms, and is determined in advance so that appropriate technical terms can be extracted. Note that this clue word is actually information indicating the clue word, but is simply referred to as a clue word. The clue word storage means 1021 can be realized by a predetermined recording medium (for example, a semiconductor memory, a magnetic disk, an optical disk, etc.). The process of storing the clue word / phrase in the clue word / phrase storage unit 1021 is not limited. For example, a clue word / phrase may be stored in the clue word / phrase storage means 1021 via a recording medium, and a clue word / phrase transmitted via a communication line or the like is stored in the clue word / phrase storage means 1021. Alternatively, the clue word / phrase input via the input device may be stored in the clue word / phrase storage unit 1021.

  The clue phrase search unit 1022 searches for clue phrases stored in the clue phrase storage unit 1021 in all or part of the document information acquired by the document information acquisition unit 1015.

  The technical term extraction unit 1023 extracts a term having a predetermined relationship with the clue phrase searched by the clue phrase search unit 1022 as a technical term. The term having a predetermined relationship with the clue word / phrase is, for example, a term before the clue word / phrase or a term after the clue word / phrase. This relationship may be determined for each clue word, or may be determined for each attribute of the clue word. The attribute of the clue word / phrase may be, for example, a relationship in which the clue word / phrase relates the technical term to the main topic. For example, there are “Method”, “Goal”, and the like as attributes of clue words. Examples of clue words having the attribute “Method” include “by”, “based on”, “using”, and the like. Examples of clue words having the attribute “Goal” include “for” and “toward”. The technical term extraction means 1023 may temporarily store the extracted technical terms on a recording medium (not shown).

  The technical term extraction unit 1023 may extract, for example, a continuous noun phrase having a predetermined relationship with a clue word phrase as a technical term, or hold a candidate word that can be a technical term in advance on a recording medium (not shown). Alternatively, terms that have a predetermined relationship with a clue word and that match a technical term candidate held therein may be extracted as a technical term.

Next, the operation of the technical term extraction unit 1016 in Pattern 1 will be described using the flowchart of FIG. The process of the flowchart of FIG. 23 corresponds to the technical term extraction process (step S1105) in the flowchart of FIG.
(Step S1601) The technical term extraction unit 1016 sets the counter i to 1 and sets the counter j to 1.

  (Step S1602) The clue phrase search unit 1022 searches the j-th clue phrase stored in the clue phrase storage unit 1021 in all or part of the i-th document information acquired by the document information acquisition unit 1015. If a search is made, the process proceeds to step S1603. If not, the process proceeds to step S1606.

  (Step S1603) The technical term extraction unit 1023 extracts a term having a predetermined relationship with the searched clue phrase as a technical term. The technical term extraction means 1023 may temporarily store the extracted technical terms on a recording medium (not shown).

(Step S1604) The technical term extraction unit 1016 increments the counter i by 1, and sets the counter j to 1.
(Step S1605) The technical term extraction unit 1016 determines whether or not the i-th document information acquired by the document information acquisition unit 1015 exists. If it exists, the process returns to step S1602, and if not, the series of processes for extracting the technical term ends, and the process proceeds to step S1106.

(Step S1606) The technical term extraction unit 1016 increments the counter j by 1.
(Step S1607) The technical term extraction unit 1016 determines whether or not the j-th clue phrase exists in the clue phrase storage unit 1021. If it exists, the process returns to step S1602, and if not, the process proceeds to step S1604.

In this flowchart, the case where the technical terms are extracted using only the clue words / phrases has been described. However, the technical terms may be extracted using the search terms together with the clue words / phrases. By doing so, the extraction accuracy of technical terms can be improved. For example, when a technical term is extracted from a part of document information “sentence extraction using SVM”, if the search term is “sentence extraction”, the search term “sentence extraction” and the clue phrase “use” are used. The term “SVM” associated with can be appropriately extracted. On the other hand, if the search term is “SVM”, the term to be extracted (the term before the clue word “using”) and the search term are the same, so that the term is extracted as inappropriate. Can be avoided. It should be noted that if at least the document information area including the search term searched by the search unit 1013 is acquired by the document information acquisition unit 1015, the technical term can be extracted using the search term. When the document information area including the search term searched by the unit 1013 is not acquired by the document information acquisition unit 1015 or the document information specified by the specifying unit 1014 and does not include the search term is acquired. In this case, technical terms may not be extracted using the search terms. Note that technical terms may also be extracted using search terms in the following patterns.
In this way, technical terms can be extracted.

[Pattern 2]
FIG. 24 is a block diagram illustrating a configuration of the technical term extraction unit 1016 in the pattern 2. 24, the technical term extraction unit 1016 includes a clue phrase storage unit 1021, a clue phrase search unit 1022, a technical term storage unit 1024, and a technical term extraction unit 1025. The configuration and operation other than the technical term storage unit 1024 and the technical term extraction unit 1025 are the same as those in the pattern 1, and the description thereof is omitted.

  The technical term storage means 1024 stores two or more technical terms. Here, technical terms are technical terms in one or more research fields and can be technical terms. This technical term is actually information indicating a technical term, but is simply referred to as a technical term. This technical term may be determined by a person, for example, or may be obtained mechanically from a plurality of technical literatures. In the latter case, for example, by extracting morphological analysis of technical literature, single nouns are extracted, scoring is performed to indicate the possibility of being a technical term for each single noun, and a single noun given a high score. Etc. may be acquired as technical terms. Here, as a method for assigning a score, a plurality of types such as scoring based on word-making ability and scoring based on appearance frequency are known. Further, other methods may be used. About the method of acquiring a technical term, since it is disclosed by the following some literature etc. and is conventionally known, the detailed description is abbreviate | omitted. Also, TermExtract has been released as a tool for creating a technical term list (http://gensen.dl.itc.u-tokyo.ac.jp/termextract.html).

  Literature: Hiroshi Nakagawa, “Automatic Term Recognition based on Statistics of Compound Nouns”, Terminology, Vol. 6, no. 2, p. 195-210, 2000

  Literature: Hirokazu Ohata, Hiroshi Nakagawa, “Extraction of technical terms by the number of words connected differently”, Information Processing Society of Japan Research Report, 2000-NL-136, p. 119-126

  References: Hiroshi Nakagawa, Masanori Mori, Yasuaki Yumoto, “Extraction of technical terms based on appearance frequency and connection frequency”, natural language processing, Vol. 10 No. 1, p. 27-45, January 2003

  The technical term storage unit 1024 can be realized by a predetermined recording medium (for example, a semiconductor memory, a magnetic disk, an optical disk, etc.). Moreover, the process in which a technical term is memorize | stored in the technical term storage means 1024 does not ask | require. For example, the technical term may be stored in the technical term storage unit 1024 via the recording medium, and the technical term transmitted via the communication line or the like is stored in the technical term storage unit 1024. Alternatively, the technical term input via the input device may be stored in the technical term storage unit 1024.

  The technical term extraction unit 1025 converts the technical term into the technical term when the technical term stored in the technical term storage unit 1024 is included in the term having a predetermined relationship with the clue term searched by the clue phrase searching unit 1022. Extract as In this way, technical terms other than the technical terms stored in the technical term storage means 1024 are not extracted as technical terms. Therefore, the technical term extraction means 1025 can use general nouns such as “research” and “method”, “proposal The situation of extracting non-technical terms such as “method” can be avoided.

  Next, the operation of the technical term extraction unit 1016 in pattern 2 will be described using the flowchart of FIG. The process of the flowchart of FIG. 25 corresponds to the technical term extraction process (step S1105) in the flowchart of FIG. In this flowchart, processes other than steps S1701 to S1706 are the same as those in the flowchart of FIG. 23 except that the technical term extraction unit 1023 is changed to the technical term extraction unit 1025, and a description thereof will be omitted.

(Step S1701) The technical term extraction means 1025 extracts terms having a predetermined relationship with the searched clue words / phrases.
(Step S1702) The technical term extraction means 1025 sets the counter k to 1.

  (Step S1703) The technical term extraction unit 1025 determines whether or not the term extracted in step S1701 includes the kth technical term stored in the technical term storage unit 1024. If it is included, the process proceeds to step S1704. If not, the process proceeds to step S1705.

(Step S1704) The technical term extraction means 1025 extracts the k-th technical term included in the term extracted in step S1701 as a technical term.
(Step S1705) The technical term extraction means 1025 increments the counter k by 1.

(Step S1706) The technical term extraction unit 1025 determines whether or not the k-th technical term exists in the technical term storage unit 1024. If it exists, the process returns to step S1703, and if not, the process proceeds to step S1604.
In this way, technical terms can be extracted.

[Pattern 3]
FIG. 26 is a block diagram illustrating a configuration of the technical term extraction unit 1016 in the pattern 3. The technical term extraction unit 1016 in this pattern extracts technical term candidates and selects technical terms from the technical term candidates. 26, the technical term extraction unit 1016 includes a clue phrase storage unit 1021, a clue phrase search unit 1022, a technical term candidate extraction unit 1026, and a technical term selection unit 1027. The configurations and operations other than the technical term candidate extraction unit 1026 and the technical term selection unit 1027 are the same as those in the pattern 1, and the description thereof is omitted.

  The technical term selection unit 1027 extracts a term having a predetermined relationship with the clue phrase searched by the clue phrase search unit 1022 as a technical term candidate. This technical term candidate extraction method is the same as the technical term extraction method described in patterns 1 and 2, and the description thereof is omitted.

  The technical term selection unit 1027 selects all or part of the technical term candidates extracted by the technical term candidate extraction unit 1026 as technical terms. A specific selection method will be described later in patterns 4 and 5.

  Next, the operation of the technical term extraction unit 1016 in pattern 3 will be described using the flowchart of FIG. The process of the flowchart of FIG. 27 corresponds to the technical term extraction process (step S1105) in the flowchart of FIG. In this flowchart, the processes other than steps S1801 and S1802 are the same as those in the flowchart of FIG.

  (Step S1801) The technical term candidate extraction unit 1026 extracts a term having a predetermined relationship with the searched clue phrase as a technical term candidate. The technical term candidate extraction unit 1026 may temporarily store the extracted technical term candidates on a recording medium (not shown).

  (Step S1802) The technical term selection unit 1027 selects a technical term from the technical term candidates selected by the technical term selection unit 1027. The technical term selection unit 1027 may temporarily store the selected technical term on a recording medium (not shown).

  In step S1802, the technical term selection unit 1027 does not select, for example, any technical term candidate that is not used as a main topic in any document information stored in the document information storage unit 1012. Technical term candidates used as main topics in the document information may be selected as technical terms. This is because a term selected as a technical term is considered to be likely to be treated as a main topic in any technical literature.

As a specific method, for example, a term indicating a main topic is previously collected from a plurality of document information and stored in a recording medium, and one of technical term candidates stored in the recording medium is stored. A word that matches the term indicating the main topic may be selected as the technical term, or a plurality of information stored in the document information storage unit 1012 indicates whether the technical term candidate is used as the term indicating the main topic. You may judge by searching the literature information. Whether or not a certain term is used as a term indicating the main topic in the document information can be determined as described in the search process in the search unit 1013 using the search term, for example. Specifically, when a certain term is described in the document information a predetermined number of times or more, it may be determined that the term is a main topic, or a certain term is defined in a plurality of predetermined areas (for example, The term may be determined to be the main topic. Needless to say, the determination may be made by other methods. For example, the structure of the document information may be analyzed to determine that the term existing in the area where the main topic is described is the main topic. Further, the method of selecting technical terms by the technical term selecting means 1027 is not limited to this, and other methods may be used. Other methods will be described later in patterns 4 and 5.
In this manner, the technical terms can be extracted by selecting the technical terms from the technical term candidates.

[Pattern 4]
FIG. 28 is a block diagram illustrating a configuration of the technical term extraction unit 1016 in the pattern 4. Here, an example of selection of technical terms in the technical term extraction unit 1016 of pattern 3 will be described. 28, the technical term extraction unit 1016 includes a clue phrase storage unit 1021, a clue phrase search unit 1022, a technical term candidate extraction unit 1026, a technical term selection unit 1027, and a technical term suffix storage unit 1028. . The configuration and operation other than the technical term suffix storage unit 1028 are the same as those in the pattern 3, and the description thereof is omitted.

The technical term suffix storage means 1028 stores one or more suffixes that can be taken by the technical term. The suffix stored in the technical term suffix storage means 1028 can be a technical term, and is, for example, “model”, “method”, “algorithm”, or the like. This suffix is actually information indicating a suffix, but is simply referred to as a suffix. The technical term suffix storage means 1028 can be realized by a predetermined recording medium (for example, a semiconductor memory, a magnetic disk, an optical disk, etc.). The process in which the suffix is stored in the technical term suffix storage means 1028 does not matter. For example, a suffix may be stored in the technical term suffix storage unit 1028 via a recording medium, and a suffix transmitted via a communication line or the like is stored in the technical term suffix storage unit 1028. Alternatively, the suffix input via the input device may be stored in the technical term suffix storage means 1028.
The technical term selection unit 1027 selects a technical term candidate having the suffix stored in the technical term suffix storage unit 1028 as a technical term.

The operation of the technical term extraction unit 1016 in this pattern is the same as the flowchart of FIG. FIG. 29 is a flowchart showing details of the technical term selection process (step S1802) in the flowchart of FIG.
(Step S1901) The technical term selection unit 1027 sets the counter i to 1 and sets the counter j to 1.

  (Step S1902) The technical term selection unit 1027 determines whether the i-th technical term candidate extracted by the technical term candidate extraction unit 1026 has the j-th suffix stored in the technical term suffix storage unit 1028. To do. If yes, the process advances to step S1903; otherwise, the process advances to step S1906.

(Step S1903) The technical term selection means 1027 selects the i-th technical term candidate as a technical term. The technical term selection unit 1027 may temporarily store the selected technical term on a recording medium (not shown).
(Step S1904) The technical term selection unit 1027 increments the counter i by 1, and sets the counter j to 1.

(Step S1905) The technical term selection unit 1027 determines whether or not the i-th technical term candidate exists. If it exists, the process returns to step S1902, and if it does not exist, the series of processes for selecting technical terms ends, and the process proceeds to step S1106.
(Step S1906) The technical term selection unit 1027 increments the counter j by 1.

(Step S1907) The technical term selection unit 1027 determines whether or not the j th suffix exists in the technical term suffix storage unit 1028. If it exists, the process returns to step S1902, and if not, the process proceeds to step S1904.
In this manner, the technical terms can be extracted by selecting the technical terms from the technical term candidates.

[Pattern 5]
FIG. 30 is a block diagram illustrating a configuration of the technical term extraction unit 1016 in the pattern 5. Here, an example of selection of technical terms in the technical term extraction unit 1016 of pattern 3 will be described. In FIG. 30, a technical term extraction unit 1016 includes a clue phrase storage unit 1021, a clue phrase search unit 1022, a technical term candidate extraction unit 1026, a technical term selection unit 1027, a technical term suffix storage unit 1028, and a document. Information specifying means 1029. The configuration and operation other than the document information specifying unit 1029 are the same as those in the pattern 4 except for the following description, and the description thereof is omitted.

  In Pattern 5, the document information acquisition unit 1015 also acquires all or part of the document information specified by the document information specifying unit 1029. The technical term selection unit 1027 stores at least one of technical term candidates extracted from all or a part of the group of literature information acquired by the literature information acquisition unit 1015 in the technical term suffix storage unit 1028. A predetermined technical term candidate is selected as a technical term from technical term candidates extracted from all or part of the group of document information. Here, “a group of document information” refers to one document information searched by the search unit 1013 or specified by the specifying unit 1014 (this is referred to as “document information N”), and the one document. This is a set of document information including one or more document information specified by the document information specifying means 1029 that the information and the field are the same. The “predetermined technical term candidate” selected by the technical term selection unit 1027 as the technical term is a technical term candidate extracted from all or a part of the document information N.

  The document information specifying method by the document information specifying unit 1029 is the same as the document information specifying method by the specifying unit 1014. For example, the document information specifying unit 1029 may specify the document information N and the document information having a citation relationship. Here, the citation relationship may be a relationship in which the document information N cites other document information, a relationship in which the document information N is cited in other document information, or both. A relationship including Moreover, you may specify the literature information which is a problem indication type citation format in which the citation format in the citation relationship points out the problem of the cited technical document. The specific operation of the document information specifying unit 1029 is the same as that of the specifying unit 1014 or can be easily considered from the operation of the specifying unit 1014, and detailed description thereof is omitted.

  The operation of the technical term extraction unit 1016 in this pattern is the same as the flowchart of FIG. FIG. 31 is a flowchart showing details of the technical term selection process (step S1802) in the flowchart of FIG. In this flowchart, processes other than step S11001 are the same as those in the flowchart of FIG. 29, and a description thereof will be omitted.

  (Step S11001) The technical term selection unit 1027 selects technical term candidates using the literature information identified by the literature information identification unit 1029. Details of this processing will be described later.

  Next, detailed processing of the technical term candidate selection processing (step S11001) in the flowchart of FIG. 31 will be described with reference to the flowchart of FIG.

  (Step S11101) The document information specifying unit 1029 specifies document information having the same field as the document information including the i-th technical term candidate. For example, when specifying document information that is cited in the document information including the i-th technical term candidate and whose citation format is type C, the process is performed by the process shown in the flowchart of FIG. Can be. For example, when the document information temporary storage unit 1020 includes a document ID or the like for identifying the document information, the document information specifying unit 1029 uses the document ID or the like to include the i th technical term candidate. Can be specified.

  (Step S11102) The document information acquiring unit 1015 acquires all or a part of each document information specified by the document information specifying unit 1029. This acquisition process can be performed in the same manner as the process of step S1104, for example. When all or a part of the document information specified by the document information specifying unit 1029 has already been acquired, the document information acquisition unit 1015 converts all or a part of the document information into the acquired document information. May be acquired from a recording medium in which all or a part of is stored, or may be acquired from the document information storage unit 1012.

(Step S11103) The technical term candidate extraction unit 1026 extracts technical term candidates from all or a part of the document information acquired in step S11102. This extraction process can be performed, for example, in the same manner as the process before step S1802 in the flowchart of FIG.
(Step S11104) The technical term selection means 1027 sets the counter m to 1 and sets the counter n to 1.

  (Step S11105) The technical term selection unit 1027 has the mth suffix stored in the technical term suffix storage unit 1028 as the nth technical term candidate extracted by the technical term candidate extraction unit 1026 in step S11103. Judge whether or not. If yes, the process proceeds to step S11110. If not, the process proceeds to step S11106.

(Step S11106) The technical term selection means 1027 increments the counter m by 1.
(Step S11107) The technical term selection unit 1027 determines whether or not the m-th suffix exists in the technical term suffix storage unit 1028. If it exists, the process returns to step S11105; otherwise, the process proceeds to step S11108.

(Step S11108) The technical term selection unit 1027 increments the counter n by 1, and sets the counter m to 1.
(Step S11109) The technical term selection unit 1027 determines whether there is an nth technical term candidate. If it exists, the process returns to step S11105. If it does not exist, a series of processes for selecting a technical term using the specified document information ends, and the process proceeds to step S1904.

  (Step S11110) The technical term selection means 1027 selects the i-th technical term candidate as a technical term. The technical term selection unit 1027 may temporarily store the selected technical term on a recording medium (not shown).

In the flowcharts of FIGS. 31 and 32, first, technical term candidates related to the document information N are selected, and then technical term candidates related to the document information identified as having the same field as the document information N are selected. However, this is merely an example, and as a result, selection of technical term candidates related to the document information N, selection of technical term candidates related to the document information identified as having the same field as the document information N, and If it can be performed, the method is not ask | required.
In this manner, the technical terms can be extracted by selecting the technical terms from the technical term candidates.

  In addition, although the technical term extraction part 1016 demonstrated the variation which extracts a technical term in said patterns 1-5, it cannot be overemphasized that a technical term may be extracted using methods other than the said description. The extracted technical terms are terms used in the technical literature field searched by the search unit 1013, and are preferably technical terms that are elements in the technical literature field, as described above. . As a method for extracting technical terms other than the above patterns 1 to 5, for example, the technical terms may be extracted by analyzing all or part of the literature information acquired by the literature information acquisition unit 1015. A method for extracting technical terms by analyzing all or part of the literature information will be briefly described.

  It is assumed that the technical term extraction unit 1016 holds information that associates the structure tag shown in FIG. 33 with the clue phrase. Then, it is assumed that “sentence extraction using SVM” is acquired by the document information acquisition unit 1015 as part of the document information. Then, the technical term extraction unit 1016 searches for a clue word phrase shown in FIG. 33 in a part of the obtained document information “sentence extraction using SVM”, and replaces the clue word phrase with a corresponding structure tag. That is, “SVM <Method clue word phrase = >> sentence extraction” is obtained. Next, the technical term extraction unit 1016 sandwiches the character string immediately before the structure tag (in the case of English, the character string immediately after) between the structure tags. In other words, as an analysis result, “<Method Cue word phrase => SVM </ Method> sentence extraction” is obtained. The technical term extraction unit 1016 can extract the term “SVM” corresponding to the “Method” tag from this analysis result as a technical term. In the case of the “Method” tag, the term sandwiched between the tags may be extracted. In the case of the “Goal” tag, the term sandwiched between the tag and a clue phrase of “Goal” Will be extracted (usually the term after the “Goal” tag). For example, when the document information acquisition unit 1015 acquires “sentence shortening for subtitle generation in a news program”, as a result of the structural analysis, “subtitle generation for <Goal clue word phrase = for news program </ Goal” // Goal > Sentence shortening, and the technical term extraction unit 1016 uses the term “sentence shortening” associated with the clue phrase “for” of the term “caption generation” and “Goal” sandwiched between “Goal” tags. Is extracted as a technical term. Note that structure tags other than those described in FIG. 33 may be used. However, the “Method” tag and the “Goal” tag are generally used for extracting technical terms. Further, the technical term extraction unit 1016 may extract technical terms using the search terms. For example, if the result of the analysis is “<Method Cue word phrase => SVM </ Method> sentence extraction” and the search term is “sentence extraction”, the technical term extraction unit 1016 displays the search term. The term “SVM”, which is a term corresponding to “sentence extraction” and sandwiched between “Method” tags, is extracted as a technical term. On the other hand, if the result of the analysis is “<SVM </ Method> sentence extraction using <Method clue word phrase =” and the search term is “SVM”, the term corresponding to the search term “SVM” is used. There is no term sandwiched between “Method” tags, so technical terms are not extracted. In this way, by extracting the technical terms using the search terms as well, it is possible to improve the accuracy of extracting the technical terms.

  Next, the operation of the technology trend information creation device 1001 according to this embodiment will be described using a specific example. In this specific example, it is assumed that the technology trend information creation device 1001 is a stand-alone device.

  In this specific example, it is assumed that the document information stored in the document information storage unit 1012 is as shown in FIG. In FIG. 34, a document ID for identifying a technical document, time information, and document information are associated with each other. The document information “D001” or the like is data indicating each document information, and at least a title area can be specified in each document information.

  First, it is assumed that the user inputs a search term “morpheme analysis” to the technology trend information creation apparatus 1001 by operating a keyboard, a mouse, or the like. Then, the search term “morpheme analysis” is received by the search term receiving unit 1011 (step S1101) and passed to the search unit 1013. The search unit 1013 searches for document information in which the search term “morpheme analysis” is included in the title area (step S1102). As a result of the search, the search unit 1013 acquires a document ID “001, 003, 010...” Corresponding to the document information including the search term “morpheme analysis” in the title area. The data is passed to the acquisition unit 1015.

  The specifying unit 1014 specifies the document information having the same field as the document information identified by the document ID received from the search unit 1013 (step S1103). In this specific example, in the document information identified by the document ID received from the search unit 1013, the document information cited in the type C citation format is specified. The process will be specifically described.

  First, the identifying unit 1014 acquires cited document identification information and citation information from the document information identified by the first document ID “001” received from the search unit 1013 (steps S1201 and S1202). When the document information identified by the document ID “001” is an academic paper, the specifying unit 1014 is described in the “reference document” column of the document information from the document information storage unit 1012. Bibliographic information of a document (title, date of publication, author, publication medium, etc.) is acquired as cited document identification information. The bibliographic information includes, for example, 1, 2, 3 ..., 1), 2), 3) ..., [1], [2], [3] ..., [A], [B], [C]. Since an index such as ... is attached, the index is also acquired. Next, the specifying unit 1014 performs a search using the acquired index in the document information, and specifies a sentence including the index. Then, a sentence including the index and a sentence closely related to the sentence are acquired as citation information. Here, a method for identifying a sentence closely related to the sentence including the index will be described.

It is assumed that the specifying unit 1014 holds the following two types of clue words / phrases in a recording medium (not shown) in advance.
(A) Pre-extended cue phrases In this, On this, Such, But, However ...
(B) Cue phrases for post-expansion We, we, They, they, Our, our, Inthis, Onthis, Such, Bute, However ...

  First, the specifying unit 1014 sets a sentence including an index as a candidate sentence for citation information. Then, when at least one of the preceding extension clue words / phrases is included in the first sentence of the candidate sentence, the sentence one sentence before the candidate sentence is included in the candidate sentence. The identifying unit 1014 repeats the process of expanding the candidate sentence before the first sentence of the candidate sentence does not include all the (A) pre-extension clue words. The specifying unit 1014 also includes a post-extension clue word / phrase included in the sentence following the candidate sentence (a sentence adjacent to the candidate sentence and not included in the candidate sentence). The next sentence of the candidate sentence is included in the candidate sentence. The specifying unit 1014 repeats the process of extending the candidate sentence later until the sentence following the candidate sentence does not include all the (B) post-extension clue words. The candidate sentence at the time when the process of extending the candidate sentence is completed becomes the citation information. The identifying unit 1014 acquires the citation information. Note that (A) the pre-extension cue word / phrase and (B) the post-extension cue word / phrase are examples, and if the sentence closely related to the sentence including the index can be appropriately extracted, A clue word may be used. In addition, here, attention is focused on the sentence including the index. However, in the technical document that directly refers to the cited document by the cited document identification information, the sentence including the cited document identification information is changed to the sentence including the index. It will be used instead.

  Next, the specifying unit 1014 determines whether the citation format is type C using the acquired citation information. Here, it is assumed that the specifying unit 1014 holds information that associates the type C clue word and the position of the clue word phrase illustrated in FIG. 35 with a recording medium (not illustrated). The “position of the clue word / phrase” in FIG. 35 is information indicating the number of sentences in the citation information with a sentence including the index and the cited document identification information as a reference (0). For example, in the case of the position “1 to 3” of the clue word phrase, the index or the cited document identification information from the sentence following the sentence including the index or the cited document identification information (the position of the clue word phrase = 1) in the citation information. Means three sentences from the sentence including the first sentence to the third backward sentence (cue phrase position = 3). When the position of the clue word is negative, it means a sentence before the sentence including the index and the cited document identification information. When the type C clue word is included in the sentence indicated by the position of the corresponding clue word phrase, the citation format of the citation information is determined to be type C. Specifically, in the cited information, when “However” is included in any sentence from the sentence (first sentence) to the third sentence after the sentence including the index and the cited document identification information. The citation format of the citation information is determined to be type C (step S1205).

  The identifying unit 1014 temporarily stores the cited document identification information corresponding to the citation information in the citation format determined to be type C (step S1206). For example, the cited document identification information corresponding to the index included in the citation information whose citation format is type C is temporarily stored. The identifying unit 1014 obtains the cited information and the cited document identification information for all the document IDs received from the search unit 1013 and temporarily stores the cited document identification information whose citation format is the type C ( Steps S1201 to S1209). The temporarily stored cited document identification information is transferred to the document information acquisition unit 1015.

  The document information acquisition unit 1015 acquires the title of the document information identified by the document ID “001, 003, 010...” Received from the search unit 1013 from the document information storage unit 1012 (steps S1301, S1401 to S1405). Further, the document information acquisition unit 1015 searches the document information storage unit 1012 using the cited document identification information received from the specifying unit 1014, and the title of the document information identified by the cited document identification information is not acquired. Then, the title of the document information is acquired (steps S1302, S1303, S1501 to S1506). In this way, the document information acquisition unit 1015 acquires a part of the document information (step S1104). FIG. 36 is a diagram illustrating an example of a part (title) of document information acquired in this way.

  Next, extraction of technical terms by the technical term extraction unit 1016 will be described. Here, a case where extraction is performed using the pattern 1 method will be described. Also, it is assumed that the information shown in FIG. 36 is stored in the document information temporary storage unit 1020. It is assumed that the clue word storage unit 1021 stores the clue words shown in FIG. In the clue word / phrase in FIG. 37, “$” is a symbol indicating the position of the technical term to be extracted. For example, when the clue word “by” is included in the title acquired by the document information acquisition unit 1015, the continuous noun phrases before the clue word “by” are extracted as technical terms.

  The first title “morphological analysis using context tree” shown in FIG. 36 does not include the clue phrases “by” and “based on”, so the clue phrase search means 1022 includes the first title. It is determined that those clue words are not included (steps S1601, S1602, S1606, S1607). Next, the clue phrase searching unit 1022 searches for the clue phrase “using” in the first title “morphological analysis using the context tree”. In this case, since “it used” is included in the title (step S1602), the clue word phrase “using” is searched, and the clue word phrase searching means 1022 before the clue word phrase “using” is searched. To the technical term extraction means 1023 to extract the continuous noun phrases. Then, the technical term extraction unit 1023 performs a morphological analysis on the title “morphological analysis using a context tree”, and extracts a continuous noun phrase “context tree” before the clue phrase “using” as a technical term ( In step S1603), the technical term “context tree” and the corresponding document ID “001” are stored in a recording medium (not shown). In this way, technical terms are extracted (step S1105). FIG. 38 is a diagram illustrating an example of information associating the technical terms extracted as described above with the document ID. Note that the technical term extraction unit 1016 passes the document ID among the information shown in FIG. 38 to the time information acquisition unit 1017. The technical term extraction unit 1016 passes the information shown in FIG. 38 to the technical trend information creation unit 1018.

  When receiving the document ID, the time information acquisition unit 1017 searches the document information storage unit 1012 using the document ID as a search key, and acquires the time information corresponding to the document ID (step S1106). FIG. 39 is a diagram illustrating an example of information associating time information acquired in this manner with a document ID. It is assumed that the time information acquisition unit 1017 passes the information shown in FIG. 39 to the technology trend information creation unit 1018.

  Upon receiving the information shown in FIG. 38 from the technical term extraction unit 1016 and the information shown in FIG. 39 from the time information acquisition unit 1017, the technical trend information creation unit 1018 groups the time information for each technical term. Each technical term is written vertically, the time is written on the horizontal axis, a circle is placed at the time corresponding to each technical term, and a line connecting the rightmost circle and the leftmost circle is described. An image is created (step S1107). The search term received through a route (not shown) is also described in the image. In order to generate an image of technical trend information, the technical trend information creation unit 1018 holds a template or the like in a recording medium (not shown), and generates an image of technical trend information using the template or the like. Good.

  The output unit 1019 displays an image of the technology trend information created by the technology trend information creation unit 1018 on the display (step S1108). FIG. 40 is a diagram showing an example of the image of the technical trend information displayed on the display in this way. In this way, the user can know the technical trend (technical flow) regarding morphological analysis.

  In this specific example, the case where the technology trend information is an image has been described. However, the technology trend information may not be an image. For example, the information shown in FIG. 41 may be combined with the information shown in FIG. 38 and the information shown in FIG. 39 using the document ID. The technical trend information shown in FIG. 41 may be spreadsheet data, for example.

  Further, in this specific example, as described in Pattern 2, technical terms may be extracted using technical terms. For example, the technical term storage means 1024 stores the technical terms shown in FIG. 42, and the terms having a predetermined relationship with the clue phrases searched by the clue phrase search means 1022 are “Japanese corpus” and “proposed method”. Suppose that In this case, since the “Japanese corpus” includes the “corpus” stored in the technical term storage unit 1024, the technical term extraction unit 1025 extracts “corpus” as the technical term. Further, since the “proposed method” does not include the “corpus” stored in the technical term storage unit 1024, the technical term extracting unit 1025 does not extract the “proposed method” that is a non-technical term as a technical term. It will be. By doing in this way, the technical term extraction means 1025 does not extract “Japanese corpus” as a technical term, but extracts an appropriate “corpus” as a technical term. Further, it is possible to avoid a situation in which the technical term extraction means 1025 extracts general nouns, non-technical terms, and the like.

  Also, the specific information used in this specific example is shown to explain the details of the operation of the technical trend information creation apparatus 1001 according to the present embodiment, and is from a corpus including actual academic papers and the like. It was not acquired.

  As described above, the technical trend information creation device 1001 according to the present embodiment extracts technical terms from the literature information, acquires time information corresponding to the literature information from which the technical terms are extracted, and extracts the extracted technology. Technology trend information can be created using terms and acquired time information. As a result, by using the created technology trend information, it is possible to easily know the transition of technology used in a certain research field. For example, it is possible to know the range of time when a technical term is described in the technical literature (from when to when the technical term is used). Further, for example, it is possible to know the frequency at which a certain technical term is described in the technical literature. In addition, there is an advantage that it is not necessary to spend a great deal of labor and time for those who are familiar with the research field in order to create the technology trend information.

  In the above-described embodiment, the case where the document information acquisition unit 1015 acquires all or a part of the document information searched by the search unit 1013 has been described. For example, the document information acquisition unit 1015 may acquire all or a part of at least some document information among the plurality of document information stored in the document information storage unit 1012. For example, when only document information of a certain research field is stored in the document information storage unit 1012, the document information acquisition unit 1015 may select all or all of the document information stored in the document information storage unit 1012. A predetermined number of documents may be acquired. As described above, when the search by the search unit 1013 is unnecessary, the technology trend information creation device 1001 does not need to include the search term reception unit 1011 or the search unit 1013.

In the above embodiment, the specifying unit 1014 specifies the document information having the same field as the document information searched by the searching unit 1013, and the document information acquiring unit 1015 specifies all or one of the document information specified by the specifying unit 1014. Although the case where the part is also acquired has been described, it may not be so. That is, the document information acquisition unit 1015 may not acquire the document information specified by the specifying unit 1014. In this case, the technology trend information creation device 1001 may not include the specifying unit 1014.
In addition, when the cited part identification information and the cited information are not used for identifying the document information by the identifying unit 1014, the document information may not include the cited document identifying information or the cited information.

(Embodiment 5)
A technical trend information creation device according to Embodiment 5 of the present invention will be described with reference to the drawings. In the technical trend information creation device according to the present embodiment, a search term accepted by the search term acceptance unit is selected.

  FIG. 43 is a block diagram showing a configuration of a technology trend information creation device 1002 according to this embodiment. 43, the technical trend information creation device 1002 according to the present embodiment includes a search term receiving unit 1011, a document information storage unit 1012, a search unit 1013, a specifying unit 1014, a document information acquisition unit 1015, and technical terms Extraction unit 1016, time information acquisition unit 1017, technology trend information creation unit 1018, output unit 1019, suffix storage unit 1031, technical term extraction unit 1032, search term storage unit 1033, search term storage unit 1034 and a search term selection receiving unit 1035. Configurations and operations other than the suffix storage unit 1031, the technical term extraction unit 1032, the search term storage unit 1033, the search term storage unit 1034, and the search term selection reception unit 1035 that are newly added in the present embodiment are accepted as a search term Unit 1011 is the same as Embodiment 4 except that it accepts the search term indicated by the selection accepted by search term selection accepting unit 1035, and a description thereof will be omitted.

  The suffix storage unit 1031 stores a predetermined suffix. This suffix is used when a search term is selected from technical terms extracted by the technical term extraction unit 1032 described later, and is, for example, “system”, “method”, “data”, or the like. Needless to say, other appropriate suffixes may be used. The suffix stored in the suffix storage unit 1031 is actually information indicating a suffix, but is simply referred to as a suffix. The suffix storage unit 1031 can be realized by a predetermined recording medium (for example, a semiconductor memory, a magnetic disk, an optical disk, etc.). Moreover, the process in which a suffix is memorize | stored in the suffix memory | storage part 1031 does not ask | require. For example, a suffix may be stored in the suffix storage unit 1031 via a recording medium, and a suffix transmitted via a communication line or the like is stored in the suffix storage unit 1031. Alternatively, the suffix input via the input device may be stored in the suffix storage unit 1031.

  The technical term extraction unit 1032 extracts technical terms from the literature information stored in the literature information storage unit 1012. Here, technical terms are technical terms in one or more research fields. This technical term is actually information indicating a technical term, but is simply referred to as a technical term. The technical term extraction unit 1032 extracts, for example, single nouns by performing morphological analysis on the literature information, performs scoring indicating the possibility of being a technical term for each single noun, and the single term given a high score. Nouns etc. may be acquired as technical terms. Here, as a method for assigning a score, a plurality of types such as scoring based on word-making ability and scoring based on appearance frequency are known. Further, other methods may be used. The method of extracting technical terms is disclosed in a plurality of documents shown in the fourth embodiment and the like, and has been conventionally known, and thus detailed description thereof is omitted.

  The search term accumulating unit 1033 adds the suffix stored in the suffix storage unit 1031 to the technical term extracted by the technical term extraction unit 1032 in the document information stored in the document information storage unit 1012. If included, the technical term is stored in the search term storage unit 1034 as a search term. By appropriately selecting a suffix stored in the suffix storage unit 1031, it becomes possible to appropriately accumulate search terms used to determine a field for creating a technical trend. For example, when the suffix “system” is stored in the suffix storage unit 1031 and the technical term extraction unit 1032 extracts “morpheme” and “morpheme analysis”, the “morpheme system” is included in the literature information. If the “morphological analysis system” is included in the document information, “morpheme analysis” is stored as a search term, but “morpheme” is not stored.

  The search term storage unit 1034 stores two or more search terms. This search term is accumulated by the search term accumulation unit 1033. The search term storage unit 1034 can be realized by a predetermined recording medium (for example, a semiconductor memory, a magnetic disk, an optical disk, etc.).

  The search term selection accepting unit 1035 accepts selection of any search term from two or more search terms stored in the search term storage unit 1034. In order to accept this selected term, two or more search terms stored in the search term storage unit 1034 may or may not be output. In the former case, for example, a search term may be displayed on a display or the like, and the user may click a desired search term by operating a mouse or the like, and the search term selection receiving unit 1035 may receive the operation. In the latter case, for example, the user has a document or the like on which two or more search terms stored in the search term storage unit 1034 and an identification symbol corresponding to each search term are displayed. An identification symbol corresponding to a desired search term may be input by operating a keyboard or the like, and the search term selection receiving unit 1035 may receive the input. The search term receiving unit 1011 receives the search term indicated by the selection received by the search term selection receiving unit 1035 as described above.

  The acceptance by the search term selection acceptance unit 1035 may be, for example, acceptance of information input from an input device (for example, a keyboard, a mouse, a touch panel, etc.), and information received via a wired or wireless communication line may be accepted. It may be received. Note that the search term selection receiving unit 1035 may or may not include a device (for example, a modem or a network card) for receiving. Further, the search term selection receiving unit 1035 may be realized by hardware, or may be realized by software such as a driver that drives a predetermined device.

  Next, the operation for creating the technology trend information in the technology trend information creating apparatus 1002 according to the present embodiment is the same as that in the flowchart of FIG. 17 except for the search term acceptance process (step S1101), and the description thereof will be omitted. FIG. 44 is a flowchart showing processing that replaces the search term acceptance processing (step S1101) in the flowchart of FIG. That is, in the technology trend information creation device 1002 according to the present embodiment, the processing of steps S111201 to S11203 shown in FIG. 44 may be performed instead of the processing of step S1101.

  (Step S11201) The search term selection receiving unit 1035 determines whether selection of a search term has been received. If the selection is accepted, the process proceeds to step S11202. If not, the process of step S11201 is repeated until the selection is accepted.

  (Step S11202) The search term selection receiving unit 1035 reads a search term indicated by the selection received by the search term selection receiving unit 1035 from two or more search terms stored in the search term storage unit 1034.

  (Step S 11203) The search term selection receiving unit 1035 inputs the search term read from the search term storage unit 1034 in Step S 11202 to the search term receiving unit 1011. As a result, the search term accepting unit 1011 accepts the search term. Then, the process proceeds to step S1102.

  In the flowchart of FIG. 44, the case where the search term selection receiving unit 1035 reads the search term stored in the search term storage unit 1034 and inputs the search term to the search term reception unit 1011 has been described. However, the search term indicated by the selection accepted by the search term selection accepting unit 1035 may be read from the search term storage unit 1034.

Next, processing for storing search terms in the search term storage unit 1034 will be described with reference to the flowchart of FIG.
(Step S11301) The technical term extraction unit 1032 extracts technical terms from the literature information stored in the literature information storage unit 1012. The method of extracting this technical term is as described above, and detailed description thereof is omitted. The extracted technical terms may be temporarily stored in a recording medium (not shown).

(Step S11302) The search term storage unit 1033 sets the counter i to 1 and sets the counter j to 1.
(Step S11303) The search term accumulation unit 1033 adds a term obtained by adding the j-th suffix stored in the suffix storage unit 1031 to the i-th technical term extracted by the technical term extraction unit 1032 as a document information storage unit. A search is performed in 1012, and it is determined whether or not the term is included in the document information stored in the document information storage unit 1012. If it is included, the process proceeds to step S11304. If not, the process proceeds to step S11307.

(Step S11304) The search term storage unit 1033 stores the i-th technical term as a search term in the search term storage unit 1034.
(Step S11305) The search term storage unit 1033 increments the counter i by 1, and sets the counter j to 1.

  (Step S11306) The search term accumulation unit 1033 determines whether or not the i-th technical term extracted by the technical term extraction unit 1032 exists. If it exists, the process returns to step S11303, and if not, the series of processes for accumulating the search terms ends.

(Step S11307) The search term storage unit 1033 increments the counter j by 1.
(Step S11308) The search term storage unit 1033 determines whether or not the j-th suffix is stored in the suffix storage unit 1031. If it is stored, the process returns to step S11303; otherwise, the process proceeds to step S11305.

  As described above, the technical trend information creation apparatus 1002 according to the present embodiment extracts technical terms, selects search terms from the technical terms, and accumulates them, thereby extracting the technical terms and selecting search terms. As a result, the search term receiving unit 1011 can receive an appropriate search term indicating the research field. As a result, appropriate technical trend information can be created.

  In this embodiment, the case where the search term storage unit 1033 stores the search terms stored in the search term storage unit 1034 has been described. However, the search terms stored in the search term storage unit 1034 may be other types. It may be accumulated by the method. For example, the user or the like may select an appropriate search term and the search term may be stored in the search term storage unit 1034. As such, the process in which the search term is stored in the search term storage unit 1034 is not limited. For example, a search term may be stored in the search term storage unit 1034 via a recording medium, and a search term transmitted via a communication line or the like is stored in the search term storage unit 1034. Alternatively, the search term input via the input device may be stored in the search term storage unit 1034. In such a case, the technology trend information creation device 1002 may not include the suffix storage unit 1031, the technical term extraction unit 1032, and the search term accumulation unit 1033.

(Embodiment 6)
A technical trend information creation device according to Embodiment 6 of the present invention will be described with reference to the drawings. The technical trend information creation device according to the present embodiment displays information related to the literature information in the displayed technical trend information image in accordance with the selection of the literature graphic corresponding to the literature information.

  FIG. 46 is a block diagram showing a configuration of a technology trend information creation device 1003 according to this embodiment. 46, the technical trend information creation device 1003 according to the present embodiment includes a search term receiving unit 1011, a document information storage unit 1012, a search unit 1013, a specifying unit 1014, a document information acquisition unit 1015, and a technical term. An extraction unit 1016, a time information acquisition unit 1017, a technology trend information creation unit 1018, an output unit 1019, a graphic selection reception unit 1041, and a display information acquisition unit 1042 are provided. The configuration and operation other than the graphic selection receiving unit 1041 and the display information acquisition unit 1042 newly added in the present embodiment are the technical trends in which the technical trend information creation unit 1018 displays the display information acquired by the display information acquisition unit 1042. Except for creating information, this is the same as in the fourth embodiment, and a description thereof is omitted.

  In the present embodiment, for each technical term, the technical trend information is displayed according to the time indicated by the time information indicated by the time information corresponding to the document information. It is assumed that this is an image to be processed. The shape of the document figure does not matter. For example, as shown in the technical trend information shown in FIG. 40, the document figure may be a circle, a triangle, a quadrangle, etc., or another shape.

  The figure selection receiving unit 1041 receives selection of a document figure in the technology trend information. For example, the selection may be accepted by using a GUI (Graphical User Interface) such as bringing the pointer over a document figure by operating the mouse or the like (on mouse over). In addition, in the technical trend information, each document graphic is associated with identification information (for example, a number or alphabet) for identifying the document graphic, and the identification is performed by the user operating the keyboard or the like. Selection may be accepted such as inputting information.

  The reception by the graphic selection receiving unit 1041 may be, for example, reception of information input from an input device (for example, a keyboard, mouse, touch panel, etc.), or reception of information transmitted via a wired or wireless communication line. But you can. The graphic selection receiving unit 1041 may or may not include a device (for example, a modem or a network card) for receiving. The graphic selection receiving unit 1041 may be realized by hardware, or may be realized by software such as a driver that drives a predetermined device.

  The display information acquisition unit 1042 acquires display information from the document information storage unit 1012. Here, the display information is information related to document information corresponding to the document graphic accepted by the graphic selection receiving unit 1041 and is information displayed on the image of the technical trend information. The information related to document information is, for example, bibliographic information of document information. More specifically, it may be the title of the document information, the author, the publication year of the document, or the like. The technical trend information creation unit 1018 creates technical trend information for displaying the display information acquired by the display information acquisition unit 1042. The display information may be displayed, for example, in the vicinity of a document graphic that has been selected by the graphic selection receiving unit 1041. Alternatively, if an area for displaying the display information is set, the display information is displayed in that area. May be.

  Next, the operation of creating the technology trend information in the technology trend information creating apparatus 1003 according to the present embodiment is the same as the flowchart of FIG. FIG. 47 is a diagram showing a process executed after the process of the technical trend information output process (step S1108) in the flowchart of FIG. That is, in the technology trend information creation device 1003 according to the present embodiment, the processing of steps S11401 to S11406 shown in FIG. 47 may be performed after the processing of step S1108.

  (Step S11401) The graphic selection receiving unit 1041 determines whether or not selection of a document graphic has been received. If the selection of the document graphic is accepted, the process proceeds to step S11402. If not, the process proceeds to step S11405.

  (Step S11402) The display information acquisition unit 1042 acquires display information, which is information related to the document graphic accepted by the graphic selection receiving unit 1041 from the document information storage unit 1012.

(Step S11403) The technology trend information creation unit 1018 creates technology trend information including the display information acquired by the display information acquisition unit 1042.
(Step S11404) The output unit 1019 outputs the technology trend information including the display information created by the technology trend information creation unit 1018. Then, the process returns to step S11401.

  (Step S11405) The output unit 1019 determines whether or not to end the output of the technology trend information. For example, output when a predetermined time has passed since the output of technology trend information, an instruction to end the output of technology trend information, or an instruction to output other information The unit 1019 may determine to end the output of the technology trend information. If the output of the technology trend information is to end, the process proceeds to step S11406. If not, the process returns to step S11401.

  (Step S11406) The output unit 1019 ends the output of the technology trend information. Then, the process returns to step S1101.

  Next, the operation of displaying the display information of the technology trend information creation apparatus 1003 according to this embodiment will be specifically described. Here, as shown in FIG. 40, it is assumed that an image of the technology trend information is displayed on the display. In the display, it is assumed that the mouse pointer is placed on the leftmost circle among the document figures (circles) corresponding to the minimum cost method. Then, the graphic selection receiving unit 1041 determines that the selection of the document graphic has been received (step S11401). Then, the graphic selection receiving unit 1041 acquires the document ID “003” for identifying the document information corresponding to the document graphic at the position of the mouse pointer by accessing the technology trend information creating unit 1018 and acquires the document ID “003” in the display information acquiring unit 1042. hand over.

  The display information acquisition unit 1042 acquires the title of the document information identified by the document ID “003” received from the graphic selection reception unit 1041 as display information from the document information storage unit 1012 (step S11402), and displays the display information. It is passed to the technology trend information creation unit 1018. The technology trend information creation unit 1018 creates the technology trend information so that the display information received from the display information acquisition unit 1042 is displayed in association with the position of the mouse pointer at that time (step S11403). The technical trend information creation unit 1018 may receive the position of the mouse pointer from the graphic selection receiving unit 1041 or may be received from another component.

  The output unit 1019 outputs the technology trend information including the display information created as described above. As a result, the technology trend information shown in FIG. 48 is displayed on the display (step S11404).

  When the user performs an operation to close the image of the technology trend information, the output unit 1019 determines that the output of the technology trend information is finished (step S11405), and the output of the technology trend information is finished. (Step S11406).

  In this specific example, the case where the title of the document information is displayed as the display information has been described. However, as described above, other display information may be displayed. For example, the display information may include an author name, an announcement time, and the like. The announcement time may be time information.

  As described above, in the technology trend information creation device 1003 according to the present embodiment, when a document graphic in an image of the technology trend information is selected, display information that is information related to document information corresponding to the document graphic is acquired. Then, an image of technical trend information including the display information is created and output. Therefore, for example, the user who is viewing the technical trend information image can easily know the predetermined information about the document information including the technical term displayed in the technical trend information image, User convenience is improved.

  Here, the improvement of the method for extracting technical terms and the experimental results will be described. In the fourth embodiment, various methods for extracting technical terms have been described. Here, methods (method 1 and method 2) in which the extraction accuracy is improved by incorporating machine learning or the like will be described.

  First, the technical term extraction unit 1016 adds a “Method” tag and a “Goal” tag to at least part of the document information acquired by the document information acquisition unit 1015 in the same manner as described above. In addition, a “Restrict” tag corresponding to a clue word related to limitation such as “in” or “related”, a “Conj” tag corresponding to a clue word related to connection such as “and”, “or”, and the like are also given. Note that tags other than the “Method” tag, “Goal” tag, and “Head” tag described later are referred to as “Other” tags. Here, a case where the document information acquisition unit 1015 acquires the title of the technical document as part of the document information will be described. In addition, this time, a “Head” tag indicating the subject of the document information is also given. The “Head” tag is attached to a noun phrase to which the “Method” tag and the “Goal” tag are not attached. In many cases, a “Head” tag is given to the noun phrase at the end of the acquired title.

  Here, a method for extracting a technical term using a title assigned with a “Head” tag or the like will be briefly described. The technical term extraction unit 1016 extracts, as a technical term, a term to which a “Method” tag is assigned from a title to which a “Head” tag is assigned to a search term received by the search term reception unit 1011. Further, the technical term extraction unit 1016 extracts, as a technical term, a term to which the “Head” tag is assigned from the title to which the “Goal” tag is assigned to the search term received by the search term reception unit 1011. By doing in this way, the technical term which shows the elemental technique with respect to a search term can be extracted.

  However, it may be inappropriate to add the “Head” tag to the noun phrase at the end of the title acquired by the document information acquisition unit 1015. For example, when the title is “Comparison experiment of SVM and minimum cost method for Chinese morphological analysis”, “Comparison experiment” at the end of the title is not the subject, but “SVM” or “Minimum cost method” is the subject. It is thought that. Therefore, even in such a case, a method (method 1 and method 2) that can appropriately add the “Head” tag will be described.

  First, method 1 will be described. In Method 1, terms inappropriate for the subject (for example, “research” “examination” “development” “impact” “analysis” “one consideration” “evaluation” “consideration” “experimental research” “one examination” “application” A list of “effects”, “suggestions”, etc.) (herein called “unnecessary word list”) is prepared. A method of creating this unnecessary word list will be described later. First, the technical term extraction unit 1016 adds a “Head” tag, a “Method” tag, and the like to the title acquired by the document information acquisition unit 1015. Then, it is determined whether the term to which the “Head” tag is added is included in the unnecessary word list. If the term to which the “Head” tag is attached is not included in the unnecessary word list, the process of assigning the “Head” tag ends. On the other hand, if the term with the “Head” tag is included in the unnecessary word list, the “Head” tag is given to the term, and the noun phrase exists before the term, A “Head” tag is assigned to a noun phrase to which no “Method” tag or “Goal” tag is attached. This process is repeated until the term with the “Head” tag is not included in the unnecessary word list. In this way, an appropriate “Head” tag is assigned. Here, the case of processing for Japanese was explained, but in the case of a language whose grammatical order is different from Japanese such as English, the term with the “Head” tag is included in the unnecessary word list. In this case, a “Head” tag is newly added to a noun phrase existing after the term.

  Next, method 2 will be described. In Method 2, data (teacher data) to which an appropriate “Head” tag is assigned is prepared and machine learning is performed. In this case, machine learning is similarly performed for “Method” tags and “Goal” tags other than the “Head” tag. In addition, by including information on whether or not each word is included in the unnecessary word list in the teacher data, it is considered that the “Head” tag is not attached to the term included in the unnecessary word list. It is done. For example, SVM (support vector machine) may be used as a machine learning tool. By performing such machine learning, it is considered that an inappropriate “Head” tag is not given.

  Next, a method for creating an unnecessary word list used in Method 1 will be described. When a “Head” tag or the like is formally added, an “Head” tag is added to an inappropriate term such as “experiment” or “one consideration” and an appropriate “morphological analysis” or the like. The term “Head” tag is also attached to the term. However, the “Head” tag is assigned to a term such as “morphological analysis” only for a paper in the natural language processing field, but the “Head” tag is assigned to a term such as “experiment” or “one consideration”. Is considered to cover a wide range of technical fields. Therefore, if the terms with the “Head” tag are sorted in order of frequency after the “Head” tag is formally assigned, unnecessary words such as “experiment” and “one consideration” are ranked higher. Become. An unnecessary word list can be created by manually excluding technical terms from unnecessary words ranked higher.

  Finally, experimental results when the “Head” tag or the like is formally added, when the method 1 is used, and when the method 2 is used will be described. In this experiment, data of NTCIR workshop 1, cross-language search task was used. This data is an abstract database from 1988 to 1997, and contains about 450,000 papers presented by 65 academic societies in Japan. An experiment was conducted by manually assigning “Head” tags and the like to 1000 Japanese paper titles randomly extracted from these data and using them as teacher data.

FIG. 49 shows the results when tags are formally added, FIG. 50 shows the results when Method 1 is used, and FIG. 51 shows the results when Method 2 is used. In these results, Precision is a value indicating accuracy (accuracy) and serves as an index of accuracy. It is obtained by the following formula.
Precision = number of morphemes whose system output and correct tag match / number of morphemes to which the system added a tag

In addition, “Recall” is a value indicating a recall rate and serves as an index of completeness. It is obtained by the following formula.
Recall = the number of morphemes for which the output of the system matches the correct tag / the number of morphemes for which a human subject gave a tag

The F value is a harmonic average of Precision and Recall, and is obtained by the following equation.
F value = 2 * Recall * Precision / (Recall + Precision)

  As can be seen from the experimental results of FIGS. 49 to 51, when a formal tag is added, for example, the extraction accuracy of the “Head” tag is a very low value of about 44%. On the other hand, in the results of the methods 1 and 2 (FIGS. 50 and 51), it can be seen that the extraction accuracy and recall of the “Head” tag are improved by 30% or more. From this, the effectiveness of Method 1 and Method 2 was confirmed. Therefore, it is possible to extract technical terms with higher accuracy by attaching tags to at least a part of the document information acquired by the document information acquisition unit 1015 using these methods 1 and 2. Become.

  Here, the improvement regarding the display of the technical trend information will be described. In the fourth embodiment and the like, the technical trend information has been described. However, when a large number of technical terms are included in the technical trend information, the technical trend information may be difficult for the user to understand. Therefore, here, a method of creating technical trend information by collecting technical terms for each classification will be described.

  FIG. 52 is a block diagram illustrating a configuration of a technology trend information creation device 1004 that creates technology trend information in which technical terms are grouped for each classification. In FIG. 52, the technology trend information creation device 1004 further includes a classification result information storage unit 1051, except that the technology trend information is created using the classification result information stored in the classification result information storage unit 1051. This is the same as the technology trend information creation device 1001 of the fourth embodiment.

  In the classification result information storage unit 1051, the classification result information is stored. Here, the classification result information is information indicating a result of classification of technical terms. This technical term is a technical term that the technical term extraction unit 1016 may extract. For example, as shown in FIG. 53, in the classification result information, technical terms are grouped for each classification. In FIG. 53, a classification name such as “algorithm” is assigned to the classification, but the classification name may not be assigned. Further, this classification result information may be created manually or automatically.

  The classification result information storage unit 1051 can be realized by a predetermined recording medium (for example, a semiconductor memory, a magnetic disk, an optical disk, etc.). The process in which the classification result information is stored in the classification result information storage unit 1051 does not matter. For example, the classification result information may be stored in the classification result information storage unit 1051 via a recording medium, and the classification result information transmitted via a communication line or the like is stored in the classification result information storage unit 1051. Alternatively, the classification result information input via the input device may be stored in the classification result information storage unit 1051.

  The technical trend information creation unit 1018 creates technical trend information in which technical terms are classified, using the classification result information stored in the classification result information storage unit 1051. Except for the technical terms being classified, the technical trend information is the same as in the fourth embodiment. Specifically, the technical trend information creation unit 1018 refers to the classification result information, and classifies the technical terms extracted by the technical term extraction unit 1016 for each classification. For example, when the technical terms “cost minimum method”, “neural network”, “HMM”, “tagged corpus”, and “untagged corpus” are extracted, the technical trend information creation unit 1018 displays the classification result information shown in FIG. The technical terms are classified into a classification “algorithm” and a classification “data”. As such, for each group of technical terms that have been classified, the technical trend information creation unit 1018 creates technical trend information in the same manner as described above. FIG. 54 is a diagram showing an example of the technical trend information created as described above. In FIG. 54, since the technical terms are divided for each classification, the user can easily grasp the technical trend for each classification. This is particularly effective when a large number of technical terms are extracted. Note that technical term information may be classified using classification result information after creating technical trend information in the same manner as in the fourth embodiment.

  In each of the above embodiments, each processing or each function may be realized by centralized processing by a single device or a single system, or distributed processing by a plurality of devices or a plurality of systems. May be realized.

  In each of the above embodiments, the related term acquisition device and the technical trend information creation device may be a stand-alone device, or may be a server device constituting a server / client system. In the latter case, the related term acquisition device may receive the term information transmitted from the client, and may transmit the related term information to the client.

  In each of the above embodiments, each component may be configured by dedicated hardware, or a component that can be realized by software may be realized by executing a program. For example, each component can be realized by a program execution unit such as a CPU reading and executing a software program recorded on a recording medium such as a hard disk or a semiconductor memory.

  In addition, the software which implement | achieves the related vocabulary acquisition apparatus in the said Embodiment 1-3 is the following programs. That is, this program has a term information receiving step for accepting term information, which is information indicating terms, in a computer, and literature information having the term indicated by the term information received in the term information accepting step in all or part of the literature information. Are searched from two or more document information stored in a document information storage unit that stores two or more document information that is information indicating a document, and all of the retrieved document information or the retrieved document information A document information acquisition step of acquiring a part having at least cited document identification information that is a part and information for identifying other documents cited in the retrieved document information from the document information storage unit, Cited text that is information indicating the document identified by the cited document identification information included in all or part of the document information acquired in the document information acquisition step Information is searched from two or more document information stored in the document information storage unit, and a part of the retrieved cited document information corresponding to a predetermined item is searched for in the document information storage unit Information indicating a related term related to the term indicated by the term information received by the term information receiving step from a part of the cited reference information acquired in the cited reference information acquiring step and the cited reference information acquiring step acquired from The related term information acquisition step for acquiring the related term information and the related term information output step for outputting the related term information acquired in the related term information acquisition step are executed.

  Moreover, the software which implement | achieves the technical trend information creation apparatus in the said embodiment is the following programs. That is, this program stores, in a computer, at least a plurality of document information that is information indicating at least a part of technical documents, and document information that stores at least time information that is information indicating the time related to the technical documents indicated by the document information. From a plurality of document information stored in the storage unit, a document information acquisition step of acquiring all or a part of at least a part of document information, and from all or a part of the document information acquired in the document information acquisition step A technical term extraction step for extracting technical terms, a timing information acquisition step for acquiring timing information corresponding to the literature information including the technical terms extracted in the technical term extraction step, and the technical term extraction Using the technical terms extracted in the step and the time information acquired in the time information acquisition step, the technical terms and the relevant technology A technology trend information creation step for creating technology trend information, which is information for associating a period related to a technical document including terms, and an output step for outputting the technology trend information created in the technology trend information creation step. Is.

  Further, in this program, the computer further executes a search term reception step for receiving a search term, and a search step for searching the document information storage unit using the search term received in the search term reception step, In the document information acquisition step, all or part of the retrieved document information may be acquired.

  In the above program, the reception step for receiving information and the output step for outputting information do not include at least processing performed only by hardware, for example, processing performed by a modem or an interface card in the output step. .

  Further, this program may be executed by being downloaded from a server or the like, and a program recorded on a predetermined recording medium (for example, an optical disk such as a CD-ROM, a magnetic disk, a semiconductor memory, or the like) is read out. May be executed by

  Further, the computer that executes this program may be singular or plural. That is, centralized processing may be performed, or distributed processing may be performed.

  FIG. 55 is a schematic diagram showing an example of the appearance of a computer that executes the program and realizes the related term acquisition device and the technical trend information creation device according to the embodiment. The above-described embodiment is realized by computer hardware and a computer program executed on the computer hardware.

  55, a computer system 100 includes a computer 101 including a CD-ROM (Compact Disk Read Only Memory) drive 105, an FD (Flexible Disk) drive 106, a keyboard 102, a mouse 103, and a monitor 104.

  FIG. 56 is a diagram illustrating a computer system. 56, in addition to the CD-ROM drive 105 and the FD drive 106, a computer 101 includes a CPU (Central Processing Unit) 111, a ROM (Read Only Memory) 112 for storing a program such as a bootup program, A CPU (Random Access Memory) 113 that is connected to the CPU 111 and temporarily stores application program instructions and provides a temporary storage space, a hard disk 114 that stores application programs, system programs, and data, a CPU 111 and a ROM 112. Etc. to each other. The computer 101 may include a network card (not shown) that provides connection to the LAN.

  A program that causes the computer system 100 to execute the functions of the related term acquisition device and the technical trend information creation device according to the above embodiment is stored in the CD-ROM 121 or the FD 122 and stored in the CD-ROM drive 105 or the FD drive 106. It may be inserted and transferred to the hard disk 114. Instead, the program may be transmitted to the computer 101 via a network (not shown) and stored in the hard disk 114. The program is loaded into the RAM 113 at the time of execution. The program may be loaded directly from the CD-ROM 121, the FD 122, or the network.

  The program does not necessarily include an operating system (OS) or a third-party program that causes the computer 101 to execute the functions of the related term acquisition device and the technical trend information creation device according to the above embodiment. The program may include only a part of an instruction that calls an appropriate function (module) in a controlled manner and obtains a desired result. How the computer system 100 operates is well known and will not be described in detail.

  Further, the present invention is not limited to the above-described embodiment, and various modifications are possible, and it goes without saying that these are also included in the scope of the present invention.

  As described above, according to the related term acquisition device or the like according to the present invention, it is possible to acquire related term information related to the accepted term information, and it is useful as a device or the like that acquires related terms related to a certain term. is there.

  In addition, according to the technical trend information creation device or the like according to the present invention, it is possible to create technical trend information, which is information that associates technical terms with the timing of technical literature including the technical terms. This is useful in an apparatus for creating

Claims (9)

A document information storage unit that stores at least a plurality of document information that is information indicating at least a part of the technical document and time information that is information indicating a time related to the technical document indicated by the document information
A search term acceptance unit for accepting search terms;
A search unit that searches the document information storage unit using a search term received by the search term reception unit ;
Among a plurality of document information stored in the document information storage unit, and the document information acquisition unit for acquiring all or part of the retrieved document information,
A technical term extraction unit that extracts technical terms from all or part of the literature information acquired by the literature information acquisition unit;
A time information acquisition unit for acquiring time information corresponding to document information including the technical terms extracted by the technical term extraction unit from the document information storage unit;
Using the technical terms extracted by the technical term extraction unit and the timing information acquired by the timing information acquisition unit, create technical trend information that is information for associating the technical terms with the timing related to the technical literature including the technical terms Technical trend information creation department to
An output unit that outputs the technology trend information created by the technology trend information creation unit;
A specifying unit for specifying the document information having the same field as the document information searched by the search unit ,
The document information acquisition unit is a technology trend information creation device that acquires all or part of the document information specified by the specifying unit from the document information storage unit . The search unit, the search term receiving unit search terms accepted searches for document information corresponding to the technical literature, which is the main topic, technology trends information creating apparatus according to claim 1. The identifying unit obtains cited document identification information that is information for identifying another technical document cited in the document information from the document information searched by the search unit, and uses the cited document identification information. The technical trend information creation device according to claim 1 , wherein the document information is identical in field to the document information searched by the search unit. The technical term extraction unit
A clue word storage means for storing one or more clue words;
In all or a part of the document information acquired by the document information acquisition unit, a clue phrase search means for searching for the clue phrase,
The terms with the cue phrase search means searches the clues words and predetermined relations, and a technical term extraction means for extracting as a technical term, technology trends information creation according to any one of claims 1 to 3 apparatus. The technical term extraction unit
A clue word storage means for storing one or more clue words;
In all or a part of the document information acquired by the document information acquisition unit, a clue phrase search means for searching for the clue phrase,
Technical term storage means for storing two or more technical terms;
Technical term extraction means for extracting the technical terms as technical terms when the technical terms stored in the technical term storage means are included in the terms having a predetermined relationship with the clue terms searched by the clue phrase searching means The technical trend information creation device according to any one of claims 1 to 3 , further comprising: The technical term extraction unit
A clue word storage means for storing one or more clue words;
In all or a part of the document information acquired by the document information acquisition unit, a clue phrase search means for searching for the clue phrase,
Technical term candidate extraction means for extracting a term having a predetermined relationship with the clue phrase searched by the clue phrase search means as a technical term candidate;
The technical trend information creation device according to any one of claims 1 to 3 , further comprising technical term selection means for selecting all or part of technical term candidates extracted by the technical term candidate extraction means as technical terms. . The technical term extraction unit
Literature information from which the technical term candidates are extracted, and literature information identifying means for identifying literature information in the same field;
Technical term suffix storage means for storing one or more suffixes that can be taken by the technical term,
The document information acquisition unit also acquires all or part of the document information specified by the document information specifying unit,
The technical term selection means includes all or a part of one piece of literature information acquired by the literature information acquisition unit, and literature information specified by the literature information specifying means that the field is the same as the one piece of literature information. When at least one of technical term candidates extracted from all or a part of the document has a suffix stored in the suffix storage means, it is extracted from all or a part of the one document information The technical trend information creation apparatus according to claim 6 , wherein a technical term candidate is selected as a technical term. The technical trend information is an image indicating the time indicated by the time information corresponding to the document information including the technical term for each technical term,
And the output unit, the technical trend information to the display output, technology trends information generating apparatus according to any one of claims 1 to 7. The technical trend information is an image that is displayed for each technical term according to the time indicated by the time information indicated by the time information corresponding to the document information, which is a graphic corresponding to the document information including the technical term.
A figure selection receiving unit for receiving selection of a document figure in the technical trend information;
A display information acquisition unit for acquiring display information, which is information related to the document information corresponding to the document graphic for which the graphic selection receiving unit has received the selection, and is displayed on the image of the technical trend information, from the document information storage unit; Further comprising
The technical trend information creation device according to claim 8 , wherein the technical trend information creation unit creates technical trend information for displaying the display information acquired by the display information acquisition unit.

Images